SNIA Chairman David Dale opening remarks SNIA DSICON 2014
SNIA DSI Conference (DSICON), CDMI Conformance Testing and other updates
DSICON is a new event produced by SNIA targeted for IT professionals involved with data storage related topics, themes, technologies and tools spanning hardware, software, cloud, virtual and physical. In this conversation, we talk about the new DSI event, the diversity of new attendees who are attending their first SNIA event, along with other updates. Some of these updates include what is new with the SNIA Cloud Data Management Initiative (CDMI), Non Volatile Memory (think flash and SSD), SMIS, education and more. In addition to the DSICON event, SNIA also announced CDMI Cloud Interoperability Conformance Test Program is now available for cloud solution vendors and providers.
DSI, Santa Clara, CA (April 22, 2014)— The Storage Networking Industry Association (SNIA), today announced the launch of a Cloud Data Management Interface (CDMI) Conformance Test Program (CTP)that validates cloud products’ conformance to the ISO/IEC CDMI standard for cloud data interoperability(ISO catalog number ISO/IEC 17826:2012). Cloud solutions that pass the CDMI CTP offer cloud consumers assurance that the CDMI standard has been properly implemented and that data stored in any conformant implementation will be transportable to any other conformant implementation.
Here is a perspective commentary quote that I issued which was included in the SNIA Press Release.
“Today, the cloud market is crowded with a slew of vendors offering different solutions for migration, data management and security, often leaving IT customers confused about the right solution for their requirements,” said Greg Schulz, founder of StorageIO, a storage technology advisory and consulting firm. “SNIA’s CDMI Conformance Test Program is a great step forward helping IT customers, VARs or others in the industry navigate their way through the fog of cloud interoperability requirements in a streamlined fashion, not to mention laying standard routes vendors will want to adopt going forward.”
Check out the full SNIA CDMI press release announcement for the conformance testing here, as well as learn more about CDMI here.
Listen in to our podcast conversation here as we cover cloud, convergence, software defined and more about data storage.
Topics and themes discussed:
Data Storage Innovation (DSICON) Conference
Data Storage Infrastructures and Convergence
Cloud Data Management Initiative (CDMI) and Conformance Testing
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved
The following are some commands and tools for Microsoft Windows environments that are useful for storage I/O activities (among others).
Finding a Windows physical disk, SSD or storage system device name
So you may know and how to find out the more familiar Windows storage device (Solid State Device – SSD, Hard Disk Drives – HDD among others) names such as A:, B:, C:, D: as what you can view from the Windows Explorer, Computer or Admin tools.
However what if you need to find out a physical name for raw (not mounted) and mounted devices for configuration? For example, if you have a tool that wants the physical name for your C: drive that might be \\.\PhysicalDrive0\?
No worries, use the command WMIC DISKDRIVE LIST BRIEF
Need more detail about the devices beyond what is shown above?
Then use WMIC DISKDRIVE LIST or as in the above example, direct the output to a file with the results shown below (scroll to the left or right to see more detail information).
Availability BytesPerSector Capabilities CapabilityDescriptions CompressionMethod ConfigManagerErrorCode ConfigManagerUserConfig DefaultBlockSize Description DeviceID ErrorCleared ErrorDescription ErrorMethodology Index InstallDate InterfaceType LastErrorCode Manufacturer MaxBlockSize MaxMediaSize MediaLoaded MediaType MinBlockSize Model Name NeedsCleaning NumberOfMediaSupported Partitions PNPDeviceID PowerManagementCapabilities PowerManagementSupported SCSIBus SCSILogicalUnit SCSIPort SCSITargetId SectorsPerTrack Signature Size Status StatusInfo SystemName TotalCylinders TotalHeads TotalSectors TotalTracks TracksPerCylinder
512 {3, 4} {"Random Access", "Supports Writing"} 0 FALSE Disk drive \\.\PHYSICALDRIVE2 2 SCSI (Standard disk drives) TRUE Fixed hard disk media ATA ST3000DM001-1CH1 SCSI Disk Device \\.\PHYSICALDRIVE2 0 SCSI\DISK&VEN_ATA&PROD_ST3000DM001-1CH1\5&3626375C&0&000600 0 0 3 6 63 0 3000590369280 OK DBIOTEST 364801 255 5860528065 93024255 255
512 {3, 4} {"Random Access", "Supports Writing"} 0 FALSE Disk drive \\.\PHYSICALDRIVE3 3 SCSI (Standard disk drives) TRUE Fixed hard disk media SEAGATE ST600MP0034 SCSI Disk Device \\.\PHYSICALDRIVE3 0 SCSI\DISK&VEN_SEAGATE&PROD_ST600MP0034\5&3626375C&0&000A00 0 0 3 10 63 600124654080 OK DBIOTEST 72961 255 1172118465 18605055 255
512 {3, 4} {"Random Access", "Supports Writing"} 0 FALSE Disk drive \\.\PHYSICALDRIVE4 4 SCSI (Standard disk drives) TRUE Fixed hard disk media SEAGATE ST600MX0004 SCSI Disk Device \\.\PHYSICALDRIVE4 0 SCSI\DISK&VEN_SEAGATE&PROD_ST600MX0004\5&3626375C&0&000C00 0 0 3 12 63 600124654080 OK DBIOTEST 72961 255 1172118465 18605055 255
512 {3, 4} {"Random Access", "Supports Writing"} 0 FALSE Disk drive \\.\PHYSICALDRIVE1 1 SCSI (Standard disk drives) TRUE Fixed hard disk media SEAGATE ST9300603SS SCSI Disk Device \\.\PHYSICALDRIVE1 0 SCSI\DISK&VEN_SEAGATE&PROD_ST9300603SS\5&3626375C&0&000400 0 0 3 4 63 299992412160 OK DBIOTEST 36472 255 585922680 9300360 255
512 {3, 4} {"Random Access", "Supports Writing"} 0 FALSE Disk drive \\.\PHYSICALDRIVE0 0 SCSI (Standard disk drives) TRUE Fixed hard disk media VMware Virtual disk SCSI Disk Device \\.\PHYSICALDRIVE0 2 SCSI\DISK&VEN_VMWARE&PROD_VIRTUAL_DISK\5&1982005&1&000000 0 0 2 0 63 -873641784 64420392960 OK DBIOTEST 7832 255 125821080 1997160 255
Remembering (or learning) Xcopy
Some of you might be familiar with Xcopy and if not, it is a handy tool for copying files, folders and directories to local as well as networked storage. Some handy Xcopy command switches include:
/j = use un-buffered IO for large files /y = suppress prompting /c = continue if error /E = copy sub directories /H = copy hidden files /Q = quiet mode (don’t list files being copied)
In the following example the content of the Videos folder and its sub-directories (83.5GB) are copied to another destination. Note the Time /T command that is also shown which is useful for timing how long the copy takes (e.g. subtract start-time from end-time and you have elapsed time). In this example 83.5GB are copied from one place to another on the same SSD device and using the results of the Time /T command the elapsed time was about 12 minutes.
Xcopy command example
Diskpart, don’t be scared, however be careful
Ever have a Windows storage device or system that failed to boot, or a problem with a partition, volume or other issue?
How about running into a situation where you are not able to format a device that you know and can confirm is ok to erase, yet you get a message that the volume is write protected or read only?
Diskpart is handy, powerful and potentially dangerous tool if you are not careful as you could mistakenly drop a good volume or partition (e.g. the importance of having good backups). However Diskpart can be used to help repair storage devices that have boot problems, or for clearing read only attributes among other tasks. If you are prefer GUI interfaces, many of the Diskpart functions can also be done via Disk Management interface (e.g. Control Panel -> All Control Panel Items -> Administrative Tools -> Computer Management -> Storage -> Disk Management ). Note that Diskpart to do certain functions will need to be run as Administrator.
In the above example the LIST DISK command shows what disks are present (on-line or off-line) which means that you may see devices here that do not show up elsewhere. Also shown is selecting a disk and then listing partitions, selecting a partition and showing attributes. The Attribute command can be used for clearing Read Only modes should a partition become write protected.
Hint, ever have a device that was once had VMware installed on it, then you move it to Windows and try to reformat for use and get a Read Only error? If so, you will want to have a look at Diskpart and the Attribute commands. However BE CAREFULL and pay attention which disk, partition and volumes you are working with as you can easily cause a problem that would result in testing how good your backups are.
Is SATA SSD TRIM Enabled?
If you have a SATA SSD the TRIM command is a form of garbage collection that is supported with Windows 7 (SAS drives use the SCSI UNMAP). Not sure if your system has TRIM enabled? Try the following command as administrator. Note that if you see a result of "0" then TRIM is enabled while a value of "1" means that it is disabled for your system.
Want to learn more about TRIM, check out this piece from Intel as well as this Microsoft Windows item.
Having issues with collecting CPU and performance statistics?
Try the Lodctr /R command (as administrator), however read this Microsoft Tip first to learn more.
Sdelete and drive erase
Like its name implies, if you do not have this tool, you can download it here from Microsoft to not only delete files, folders, as well as write "0" patterns across a disk to secure erase it. You can specify the number of times you want to run the write "0" patterns across a disk to meet your erasure requirements.
There is also another use for Sdelete which is if you need or want to pre-condition a SSD or other device such as for testing, you can run a pre-conditioning pass using Sdelete.
Some command options include -p #n where "n" is the number of times to run, -s recursive to process sub-directories, -z to write "0" or zero out the space on the device, -c for clean, -a to process read-only attributes. Learn more and get your copy of Sdelete from Microsoft here.
Rufus, Seatools, Samsung Disk Magician and Cyberduck
A handy tool available from Seagate (may only work with Seagate and their partner devices) is SeaTools that can give drive information, health and status as well as perform various tests including SMART.
Seagate Seatools example
Different HDD and SSD as well as storage system vendors give tools for configuration, monitoring, management and in some cases data movement with their solutions. Samsung SSD Magician is a tool I have installed for managing my SSDs (830 and 840 Pros) that has features for updating firmware, drive health as well as performance optimization. Other hand tools include the Samsung copy tool based on Clonix as Acronis among other clone or data migration utilities (more on those in a future post).
Samsung SSD Magician
While the Microsoft WIndows USB Tool is handy for dealing with Microsoft ISO, however for creating USB’s with ISO’s such as for installing VMware or Linux on bare metal systems, Rufus is a handy tool to have in the tool-box.
Another useful tool that functions as an SSH and FTP utility is Cyberduck that also supports access to Amazon S3 among other cloud services.
There are many other tools for server, storage I/O and other activities on WIndows, not to mention other platforms, however hopefully you find the above useful.
How about it, what’s your favorite Windows server, storage I/O tools and commands?
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved
Seagate Kinetic Cloud and Object Storage I/O platform
Seagate announced today their Kinetic platform and drive designed for use by object API accessed storage including for cloud deployments. The Kinetic platform includes Hard Disk Drives (HDD) that feature 1Gb Ethernet (1 GbE) attached devices that speak object access API or what Seagate refers to as a key / value.
What is being announced with Seagate Kinetic Cloud and Object (Ethernet HDD) Storage?
Kinetic Open Storage Platform – Ethernet drives, key / value (object access) API, partner software
Software developer’s kits (SDK) – Developer tools, documentation, drive simulator, code libraries, code samples including for SwiftStack and Riak.
Partner ecosystem
What is Kinetic?
While it has 1 GbE ports, do not expect to be able to use those for iSCSI or NAS including NFS, CIFS or other standard access methods. Being Ethernet based, the Kinetic drive only supports the key value object access API. What this means is that applications, cloud or object stacks, key value and NoSQL data repositories, or other software that adopt the API can communicate directly using object access.
Internal, the HDD functions as a normal drive would store and accessing data, the object access function and translation layer shifts from being in an Object Storage Device (OSD) server node to inside the HDD. The Kinetic drive takes on the key value API personality over 1 GbE ports instead of traditional Logical Block Addressing (LBA) and Logical Block Number (LBN) access using 3g, 6g or emerging 12g SAS or SATA interfaces. Instead Kinetic drives respond to object access (aka what Seagate calls key / value) API commands such as Get, Put among others. Learn more about object storage, access and clouds at www.objectstoragecenter.com.
Some questions and comments
Is this the same as what was attempted almost a decade ago now with the T10 OSD drives?
Seagate claims no.
What is different this time around with Seagate doing a drive that to some may vaguely resemble the predecessor failed T10 OSD approach?
Industry support for object access and API development have progressed from an era of build it and they will come thinking, to now where the drives are adapted to support current cloud, object and key value software deployment.
Wont 1GbE ports be too slow vs. 12g or 6g or even 3g SAS and SATA ports?
Keep in mind those would be apples to oranges comparisons based on the protocols and types of activity being handled. Kinetic types of devices initially will be used for large data intensive applications where emphasis is on storing or retrieving large amounts of information, vs. low latency transactional. Also, keep in mind that one of the design premises is to keep cost low, spread the work over many nodes, devices to meet those goals while relying on server-side caching tools.
Does this mean that the HDD is actually software defined?
Seagate or other HDD manufactures have not yet noticed the software defined marketing (SDM) bandwagon. They could join the software defined fun (SDF) and talk about a software defined disk (SDD) or software defined HDD (SDHDD) however let us leave that alone for now.
The reality is that there is far more software that exists in a typical HDD than what is realized. Sure some of that is packaged inside ASICs (Application Specific Integrated Circuits) or running as firmware that can be updated. However, there is a lot of software running in a HDD hence the need for power yet energy-efficient processors found in those devices. On a drive per drive basis, you may see a Kinetic device consume more energy vs. other equivalence HDDs due to the increase in processing (compute) needed to run the extra software. However that also represents an off-load of some work from servers enabling them to be smaller or do more work.
Are these drives for everybody?
It depends on if your application, environment, platform and technology can leverage them or not. This means if you view the world only through what is new or emerging then these drives may be for all of those environments, while other environments will continue to leverage different drive options.
Does this mean that block storage access is now dead?
Not quite, after all there is still some block activity involved, it is just that they have been further abstracted. On the other hand, many applications, systems or environments still rely on block as well as file based access.
Seagate has indicated those and others are targeted to be included in the ecosystem.
Seagate needs to be careful balancing their story and message with Kinetic to play to and support those focused on the new and emerging, while also addressing their bread and butter legacy markets. The balancing act is communicating options, flexibility to choose and adopt the right technology for the task without being scared of the future, or clinging to the past, not to mention throwing the baby out with the bath water in exchange for something new.
For those looking to do object storage systems, or cloud and other scale based solutions, Kinetic represents a new tool to do your due diligence and learn more about.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved
Can we get a side of context with them server storage metrics?
Updated 2/10/2018
Whats the best server storage I/O network metric or benchmark? It depends as there needs to be some context with them IOPS and other server storage I/O metrics that matter.
In the meantime, let’s get a side of some context with them IOPS from vendors, marketers and their pundits who are tossing them around for server, storage and IO metrics that matter.
Expanding the conversation, the need for more context
The good news is that people are beginning to discuss storage beyond space capacity and cost per GByte, TByte or PByte for both DRAM or nand flash Solid State Devices (SSD), Hard Disk Drives (HDD) along with Hybrid HDD (HHDD) and Solid State Hybrid Drive (SSHD) based solutions. This applies to traditional enterprise or SMB IT data center with physical, virtual or cloud based infrastructures.
This is good because it expands the conversation beyond just cost for space capacity into other aspects including performance (IOPS, latency, bandwidth) for various workload scenarios along with availability, energy effective and management.
Adding a side of context
The catch is that IOPS while part of the equation are just one aspect of performance and by themselves without context, may have little meaning if not misleading in some situations.
Granted it can be entertaining, fun to talk about or simply make good press copy for a million IOPS. IOPS vary in size depending on the type of work being done, not to mention reads or writes, random and sequential which also have a bearing on data throughout or bandwidth (Mbytes per second) along with response time.
However, are those million IOP’s applicable to your environment or needs?
Likewise, what do those million or more IOPS represent about type of work being done? For example, are they small 64 byte or large 64 Kbyte sized, random or sequential, cached reads or lazy writes (deferred or buffered) on a SSD or HDD?
How about the response time or latency for achieving them IOPS?
In other words, what is the context of those metrics and why do they matter?
Metrics that matter give context for example IO sizes closer to what your real needs are, reads and writes, mixed workloads, random or sequential, sustained or bursty, in other words, real world reflective.
As with any benchmark take them with a grain (or more) of salt, they key is use them as an indicator then align to your needs. The tool or technology should work for you, not the other way around.
Here are some examples of context that can be added to help make IOP’s and other metrics matter:
What is the IOP size, are they 512 byte (or smaller) vs. 4K bytes (or larger)?
Are they reads, writes, random, sequential or mixed and what percentage?
How was the storage configured including RAID, replication, erasure or dispersal codes?
Then there is the latency or response time and IO queue depths for the given number of IOPS.
Let us not forget if the storage systems (and servers) were busy with other work or not.
If there is a cost per IOP, is that list price or discount (hint, if discount start negotiations from there)
What was the number of threads or workers, along with how many servers?
What tool was used, its configuration, as well as raw or cooked (aka file system) IO?
Was the IOP’s number with one worker or multiple workers on a single or multiple servers?
Did the IOP’s number come from a single storage system or total of multiple systems?
Fast storage needs fast serves and networks, what was their configuration?
Was the performance a short burst, or long sustained period?
What was the size of the test data used; did it all fit into cache?
Were write data committed synchronously to storage, or deferred (aka lazy writes used)?
The above are just a sampling and not all may be relevant to your particular needs, however they help to put IOP’s into more contexts. Another consideration around IOPS are the configuration of the environment, from an actual running application using some measurement tool, or are they generated from a workload tool such as IOmeter, IOrate, VDbench among others.
Sure, there are more contexts and information that would be interesting as well, however learning to walk before running will help prevent falling down.
Does size or age of vendors make a difference when it comes to context?
Some vendors are doing a good job of going for out of this world record-setting marketing hero numbers.
Meanwhile other vendors are doing a good job of adding context to their IOP or response time or bandwidth among other metrics that matter. There is a mix of startup and established that give context with their IOP’s or other metrics, likewise size or age does not seem to matter for those who lack context.
Some vendors may not offer metrics or information publicly, so fine, go under NDA to learn more and see if the results are applicable to your environments.
Likewise, if they do not want to provide the context, then ask some tough yet fair questions to decide if their solution is applicable for your needs.
What this means is let us start putting and asking for metrics that matter such as IOP’s with context.
If you have a great IOP metric, if you want it to matter than include some context such as what size (e.g. 4K, 8K, 16K, 32K, etc.), percentage of reads vs. writes, latency or response time, random or sequential.
IMHO the most interesting or applicable metrics that matter are those relevant to your environment and application. For example if your main application that needs SSD does about 75% reads (random) and 25% writes (sequential) with an average size of 32K, while fun to hear about, how relevant is a million 64 byte read IOPS? Likewise when looking at IOPS, pay attention to the latency, particular if SSD or performance is your main concern.
Get in the habit of asking or telling vendors or their surrogates to provide some context with them metrics if you want them to matter.
So how about some context around them IOP’s (or latency and bandwidth or availability for that matter)?
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.
This is the second post of a two-part series looking at storage performance, specifically in the context of drive or device (e.g. mediums) characteristics of How many IOPS can a HDD HHDD SSD do with VMware. In the first post the focus was around putting some context around drive or device performance with the second part looking at some workload characteristics (e.g. benchmarks).
Here are some examples to give you some more insight.
For example, the following shows how IOPS vary by changing the percent of reads, writes, random and sequential for a 4K (4,096 bytes or 4 KBytes) IO size with each test step (4 minutes each).
IO Size for test
Workload Pattern of test
Avg. Resp (R+W) ms
Avg. IOP Sec (R+W)
Bandwidth KB Sec (R+W)
4KB
100% Seq 100% Read
0.0
29,736
118,944
4KB
60% Seq 100% Read
4.2
236
947
4KB
30% Seq 100% Read
7.1
140
563
4KB
0% Seq 100% Read
10.0
100
400
4KB
100% Seq 60% Read
3.4
293
1,174
4KB
60% Seq 60% Read
7.2
138
554
4KB
30% Seq 60% Read
9.1
109
439
4KB
0% Seq 60% Read
10.9
91
366
4KB
100% Seq 30% Read
5.9
168
675
4KB
60% Seq 30% Read
9.1
109
439
4KB
30% Seq 30% Read
10.7
93
373
4KB
0% Seq 30% Read
11.5
86
346
4KB
100% Seq 0% Read
8.4
118
474
4KB
60% Seq 0% Read
13.0
76
307
4KB
30% Seq 0% Read
11.6
86
344
4KB
0% Seq 0% Read
12.1
82
330
Dell/Western Digital (WD) 1TB 7200 RPM SATA HDD (Raw IO) thread count 1 4K IO size
In the above example the drive is a 1TB 7200 RPM 3.5 inch Dell (Western Digital) 3Gb SATA device doing raw (non file system) IO. Note the high IOP rate with 100 percent sequential reads and a small IO size which might be a result of locality of reference due to drive level cache or buffering.
Some drives have larger buffers than others from a couple to 16MB (or more) of DRAM that can be used for read ahead caching. Note that this level of cache is independent of a storage system, RAID adapter or controller or other forms and levels of buffering.
Does this mean you can expect or plan on getting those levels of performance?
I would not make that assumption, and thus this serves as an example of using metrics like these in the proper context.
Building off of the previous example, the following is using the same drive however with a 16K IO size.
IO Size for test
Workload Pattern of test
Avg. Resp (R+W) ms
Avg. IOP Sec (R+W)
Bandwidth KB Sec (R+W)
16KB
100% Seq 100% Read
0.1
7,658
122,537
16KB
60% Seq 100% Read
4.7
210
3,370
16KB
30% Seq 100% Read
7.7
130
2,080
16KB
0% Seq 100% Read
10.1
98
1,580
16KB
100% Seq 60% Read
3.5
282
4,522
16KB
60% Seq 60% Read
7.7
130
2,090
16KB
30% Seq 60% Read
9.3
107
1,715
16KB
0% Seq 60% Read
11.1
90
1,443
16KB
100% Seq 30% Read
6.0
165
2,644
16KB
60% Seq 30% Read
9.2
109
1,745
16KB
30% Seq 30% Read
11.0
90
1,450
16KB
0% Seq 30% Read
11.7
85
1,364
16KB
100% Seq 0% Read
8.5
117
1,874
16KB
60% Seq 0% Read
10.9
92
1,472
16KB
30% Seq 0% Read
11.8
84
1,353
16KB
0% Seq 0% Read
12.2
81
1,310
Dell/Western Digital (WD) 1TB 7200 RPM SATA HDD (Raw IO) thread count 1 16K IO size
The previous two examples are excerpts of a series of workload simulation tests (ok, you can call them benchmarks) that I have done to collect information, as well as try some different things out.
The following is an example of the summary for each test output that includes the IO size, workload pattern (reads, writes, random, sequential), duration for each workload step, totals for reads and writes, along with averages including IOP’s, bandwidth and latency or response time.
Want to see more numbers, speeds and feeds, check out the following table which will be updated with extra results as they become available.
Performance characteristics 1 worker (thread count) for RAW IO (non-file system)
Note: (1) Seagate Momentus XT is a Hybrid Hard Disk Drive (HHDD) based on a 7.2K 2.5 HDD with SLC nand flash integrated for read buffer in addition to normal DRAM buffer. This model is a XT I (4GB SLC nand flash), may add an XT II (8GB SLC nand flash) at some future time.
As a starting point, these results are raw IO with file system based information to be added soon along with more devices. These results are for tests with one worker or thread count, other results will be added with such as 16 workers or thread counts to show how those differ.
The above results include all reads, all writes, mix of reads and writes, along with all random, sequential and mixed for each IO size. IO sizes include 4K, 8K, 16K, 32K, 64K, 128K, 256K, 512K, 1024K and 2048K. As with any workload simulation, benchmark or comparison test, take these results with a grain of salt as your mileage can and will vary. For example you will see some what I consider very high IO rates with sequential reads even without file system buffering. These results might be due to locality of reference of IO’s being resolved out of the drives DRAM cache (read ahead) which vary in size for different devices. Use the vendor model numbers in the table above to check the manufactures specs on drive DRAM and other attributes.
If you are used to seeing 4K or 8K and wonder why anybody would be interested in some of the larger sizes take a look at big fast data or cloud and object storage. For some of those applications 2048K may not seem all that big. Likewise if you are used to the larger sizes, there are still applications doing smaller sizes. Sorry for those who like 512 byte or smaller IO’s as they are not included. Note that for all of these unless indicated a 512 byte standard sector or drive format is used as opposed to emerging Advanced Format (AF) 4KB sector or block size. Watch for some more drive and device types to be added to the above, along with results for more workers or thread counts, along with file system and other scenarios.
Using VMware as part of a Server, Storage and IO (aka StorageIO) test platform
The above performance results were generated on Ubuntu 12.04 (since upgraded to 14.04 which was hosted on a VMware vSphere 5.1 (upgraded to 5.5U2) purchased version (you can get the ESXi free version here) with vCenter enabled system. I also have VMware workstation installed on some of my Windows-based laptops for doing preliminary testing of scripts and other activity prior to running them on the larger server-based VMware environment. Other VMware tools include vCenter Converter, vSphere Client and CLI. Note that other guest virtual machines (VMs) were idle during the tests (e.g. other guest VMs were quiet). You may experience different results if you ran Ubuntu native on a physical machine or with different adapters, processors and device configurations among many other variables (that was a disclaimer btw ;) ).
All of the devices (HDD, HHDD, SSD’s including those not shown or published yet) were Raw Device Mapped (RDM) to the Ubuntu VM bypassing VMware file system.
Example of creating an RDM for local SAS or SATA direct attached device.
The above uses the drives address (find by doing a ls -l /dev/disks via VMware shell command line) to then create a vmdk container stored in a dat. Note that the RDM being created does not actually store data in the .vmdk, it’s there for VMware management operations.
If you are not familiar with how to create a RDM of a local SAS or SATA device, check out this post to learn how.This is important to note in that while VMware was used as a platform to support the guest operating systems (e.g. Ubuntu or Windows), the real devices are not being mapped through or via VMware virtual drives.
The above shows examples of RDM SAS and SATA devices along with other VMware devices and dats. In the next figure is an example of a workload being run in the test environment.
One of the advantages of using VMware (or other hypervisor) with RDM’s is that I can quickly define via software commands where a device gets attached to different operating systems (e.g. the other aspect of software defined storage). This means that after a test run, I can quickly simply shutdown Ubuntu, remove the RDM device from that guests settings, move the device just tested to a Windows guest if needed and restart those VMs. All of that from where ever I happen to be working from without physically changing things or dealing with multi-boot or cabling issues.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.
How many i/o iops can flash ssd or hdd do with vmware?
Updated 2/10/2018
A common question I run across is how many I/O iopsS can flash SSD or HDD storage device or system do or give.
The answer is or should be it depends.
This is the first of a two-part series looking at storage performance, and in context specifically around drive or device (e.g. mediums) characteristics across HDD, HHDD and SSD that can be found in cloud, virtual, and legacy environments. In this first part the focus is around putting some context around drive or device performance with the second part looking at some workload characteristics (e.g. benchmarks).
Lets leave those for a different discussion at another time.
Getting started
Part of my interest in tools, metrics that matter, measurements, analyst, forecasting ties back to having been a server, storage and IO performance and capacity planning analyst when I worked in IT. Another aspect ties back to also having been a sys admin as well as business applications developer when on the IT customer side of things. This was followed by switching over to the vendor world involved with among other things competitive positioning, customer design configuration, validation, simulation and benchmarking HDD and SSD based solutions (e.g. life before becoming an analyst and advisory consultant).
Btw, if you happen to be interested in learn more about server, storage and IO performance and capacity planning, check out my first book Resilient Storage Networks (Elsevier) that has a bit of information on it. There is also coverage of metrics and planning in my two other books The Green and Virtual Data Center (CRC Press) and Cloud and Virtual Data Storage Networking (CRC Press). I have some copies of Resilient Storage Networks available at a special reader or viewer rate (essentially shipping and handling). If interested drop me a note and can fill you in on the details.
There are many rules of thumb (RUT) when it comes to metrics that matter such as IOPS, some that are older while others may be guess or measured in different ways. However the answer is that it depends on many things ranging from if a standalone hard disk drive (HDD), Hybrid HDD (HHDD), Solid State Device (SSD) or if attached to a storage system, appliance, or RAID adapter card among others.
Taking a step back, the big picture
Various HDD, HHDD and SSD’s
Server, storage and I/O performance and benchmark fundamentals
Even if just looking at a HDD, there are many variables ranging from the rotational speed or Revolutions Per Minute (RPM), interface including 1.5Gb, 3.0Gb, 6Gb or 12Gb SAS or SATA or 4Gb Fibre Channel. If simply using a RUT or number based on RPM can cause issues particular with 2.5 vs. 3.5 or enterprise and desktop. For example, some current generation 10K 2.5 HDD can deliver the same or better performance than an older generation 3.5 15K. Other drive factors (see this link for HDD fundamentals) including physical size such as 3.5 inch or 2.5 inch small form factor (SFF), enterprise or desktop or consumer, amount of drive level cache (DRAM). Space capacity of a drive can also have an impact such as if all or just a portion of a large or small capacity devices is used. Not to mention what the drive is attached to ranging from in internal SAS or SATA drive bay, USB port, or a HBA or RAID adapter card or in a storage system.
HDD fundamentals
How about benchmark and performance for marketing or comparison tricks including delayed, deferred or asynchronous writes vs. synchronous or actually committed data to devices? Lets not forget about short stroking (only using a portion of a drive for better IOP’s) or even long stroking (to get better bandwidth leveraging spiral transfers) among others.
Almost forgot, there are also thick, standard, thin and ultra thin drives in 2.5 and 3.5 inch form factors. What’s the difference? The number of platters and read write heads. Look at the following image showing various thickness 2.5 inch drives that have various numbers of platters to increase space capacity in a given density. Want to take a wild guess as to which one has the most space capacity in a given footprint? Also want to guess which type I use for removable disk based archives along with for onsite disk based backup targets (compliments my offsite cloud backups)?
Thick, thin and ultra thin devices
Beyond physical and configuration items, then there are logical configuration including the type of workload, large or small IOPS, random, sequential, reads, writes or mixed (various random, sequential, read, write, large and small IO). Other considerations include file system or raw device, number of workers or concurrent IO threads, size of the target storage space area to decide impact of any locality of reference or buffering. Some other items include how long the test or workload simulation ran for, was the device new or worn in before use among other items.
Tools and the performance toolbox
Then there are the various tools for generating IO’s or workloads along with recording metrics such as reads, writes, response time and other information. Some examples (mix of free or for fee) include Bonnie, Iometer, Iorate, IOzone, Vdbench, TPC, SPC, Microsoft ESRP, SPEC and netmist, Swifttest, Vmark, DVDstore and PCmark 7 among many others. Some are focused just on the storage system and IO path while others are application specific thus exercising servers, storage and IO paths.
Server, storage and IO performance toolbox
Having used Iometer since the late 90s, it has its place and is popular given its ease of use. Iometer is also long in the tooth and has its limits including not much if any new development, never the less, I have it in the toolbox. I also have Futremark PCmark 7 (full version) which turns out has some interesting abilities to do more than exercise an entire Windows PC. For example PCmark can use a secondary drive for doing IO to.
PCmark can be handy for spinning up with VMware (or other tools) lots of virtual Windows systems pointing to a NAS or other shared storage device doing real world type activity. Something that could be handy for testing or stressing virtual desktop infrastructures (VDI) along with other storage systems, servers and solutions. I also have Vdbench among others tools in the toolbox including Iorate which was used to drive the workloads shown below.
What I look for in a tool are how extensible are the scripting capabilities to define various workloads along with capabilities of the test engine. A nice GUI is handy which makes Iometer popular and yes there are script capabilities with Iometer. That is also where Iometer is long in the tooth compared to some of the newer generation of tools that have more emphasis on extensibility vs. ease of use interfaces. This also assumes knowing what workloads to generate vs. simply kicking off some IOPs using default settings to see what happens.
Another handy tool is for recording what’s going on with a running system including IO’s, reads, writes, bandwidth or transfers, random and sequential among other things. This is where when needed I turn to something like HiMon from HyperIO, if you have not tried it, get in touch with Tom West over at HyperIO and tell him StorageIO sent you to get a demo or trial. HiMon is what I used for doing start, stop and boot among other testing being able to see IO’s at the Windows file system level (or below) including very early in the boot or shutdown phase.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.
This is part III in a series of posts pertaining to EMC ViPR software defined storage and object storage. You can read part I here and part II here.
More on the object opportunity
Other object access includes OpenStack storage part Swift, AWS S3 HTTP and REST API access. This also includes ViPR supporting EMC Atmos, VNX and Isilon arrays as southbound persistent storage in addition.
Object (and cloud) storage access example
EMC is claiming that over 250 VNX systems can be abstracted to support scaling with stability (performance, availability, capacity, economics) using ViPR. Third party storage will be supported along with software such as OpenStack Swift, Ceph and others running on commodity hardware. Note that EMC has some history with object storage and access including Centera and Atmos. Visit the micro site I have setup called www.objectstoragecenter.com and watch for more content to be updated and added there.
More on the ViPR control plane and controller
ViPR differs from some others in that it does not sit in the data path all the time (e.g. between application servers and storage systems or cloud services) to cut potential for bottlenecks.
Organizations that can use ViPR include enterprise, SMB, CSP or MSP and hosting sites. ViPR can be used in a control mode to leverage underlying storage systems, appliances and services intelligence and functionality. This means ViPR can be used to complement as oppose to treat southbound or target storage systems and services as dumb disks or JBOD.
On the other hand, ViPR will also have a suite of data services such as snapshot, replication, data migration, movement, tiering to add value for when those do not exist. Customers will be free to choose how they want to use and deploy ViPR. For example leveraging underlying storage functionality (e.g. lightweight model), or in a more familiar storage virtualization model heavy lifting model. In the heavy lifting model more work is done by the virtualization or abstraction software to create an added value, however can be a concern for bottlenecks depending how deployed.
Software defined, storage hypervisor, virtual storage or storage virtualization?
Most storage virtualization, storage hypervisors and virtual storage solutions that are hardware or software based (e.g. software defined) implemented what is referred to as in band. With in band the storage virtualization software or hardware sits between the applications (northbound) and storage systems or services (southbound).
While this approach can be easier to carry out along with add value add services, it can also introduce scaling bottlenecks depending on implementations. Examples of in band storage virtualization includes Actifio, DataCore, EMC VMAX with third-party storage, HDS with third-party storage, IBM SVC (and their V7000 Storwize storage system based on it) and NetApp Vseries among others. An advantage of in band approaches is that there should not need to be any host or server-side software requirements and SAN transparency.
There is another approach called out-of-band that has been tried. However pure out-of-band requires a management system along with agents, drivers, shims, plugins or other software resident on host application servers.
Example of generic fast path control path model
ViPR takes a different approach, one that was seen a few years ago with EMC Invista called fast path, control path that for the most part stays out of the data path. While this is like out-of-band, there should not be a need for any host server-side (e.g. northbound) software. By being a fast path control path, the virtualization or abstraction and management functions stay out of the way for data being moved or work being done.
Hmm, kind of like how management should be, there to help when needed, out-of-the-way not causing overhead other times ;).
Is EMC the first (even with Invista) to leverage fast path control path?
Actually up until about a year or so ago, or shortly after HP acquired 3PAR they had a solution called Storage Virtualization Services Platform (SVPS) that was OEMd from LSI (e.g. StorAge). Unfortunately, HP decided to retire that as opposed to extend its capabilities for file and object access (northbound) as well as different southbound targets or destination services.
Whats this northbound and southbound stuff?
Simply put, think in terms of a vertical stack with host servers (PMs or VMs) on the top with applications (and hypervisors or other tools such as databases) on top of them (e.g. north).
Northbound servers, southbound storage systems and cloud services
Think of storage systems, appliances, cloud services or other target destinations on the bottom (or south). ViPR sits in between providing storage services and management to the northbound servers leveraging the southbound storage.
What host servers can VIPR support for serving storage?
VIPR is being designed to be server agnostic (e.g. virtual or physical), along with operating system agnostic. In addition VIPR is being positioned as capable of serving northbound (e.g. up to application servers) block, file or object as well as accessing southbound (e.g. targets) block, file and object storage systems, file systems or services.
Note that a difference between earlier similar solutions from EMC have been either block based (e.g. Invista, VPLEX, VMAX with third-party storage), or file based. Also note that this means VIPR is not just for VMware or virtual server environments and that it can exist in legacy, virtual or cloud environments.
Likewise VIPR is intended to be application agnostic supporting little data, big data, very big data ( VBD) along with Hadoop or other specialized processing. Note that while VIPR will support HDFS in addition to NFS and CIFS file based access, Hadoop will not be running on or in the VIPR controllers as that would live or run elsewhere.
How will VIPR be deployed and licensed?
EMC has indicated that the VIPR controller will be delivered as software that installs into a virtual appliance (e.g. VMware) running as a virtual machine (VM) guest. It is not clear when support will exist for other hypervisors (e.g. Microsoft Hyper-V, Citrix/XEN, KVM or if VMware vSphere with vCenter or simply on ESXi free version). As of the announcement pre briefing, EMC had not yet finalized pricing and licensing details. General availability is expected in the second half of calendar 2013.
Keep in mind that the VIPR controller (software) runs as a VM that can be hosted on a clustered hypervisor for HA. In addition, multiple VIPR controllers can exist in a cluster to further enhance HA.
Some questions to be addressed among others include:
How and where are IOs intercepted?
Who can have access to the APIs, what is the process, is there a developers program, SDK along with resources?
What network topologies are supported local and remote?
What happens when JBOD is used and no advanced data services exist?
What are the characteristics of the object access functionality?
What if any specific switches or data path devices and tools are needed?
How does a host server know to talk with its target and ViPR controller know when to intercept for handling?
Will SNIA CDMI be added and when as part of the object access and data services capabilities?
Are programmatic bindings available for the object access along with support for other APIs including IOS?
What are the performance characteristics including latency under load as well as during a failure or fault scenario?
How will EMC place Vplex and its caching model on a local and wide area basis vs. ViPR or will we see those two create some work together, if so, what will that be?
Bottom line (for now):
Good move for EMC, now let us see how they execute including driving adoption of their open APIs, something they have had success in the past with Centera and other solutions. Likewise, let us see what other storage vendors become supported or add support along with how pricing and licensing are rolled out. EMC will also have to articulate when and where to use ViPR vs. VPLEX along with other storage systems or management tools.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved
Organizations that need to scale with stability across EMC, third-party or open storage software stacks and commodity hardware. This applies to large and small enterprise, cloud service providers, managed service providers, virtual and cloud environments/
What this means for EMC hardware/platform/systems?
They can continue to be used as is, or work with ViPR or other deployment modes.
Does this mean EMC storage systems are nearing their end of life?
IMHO for the most part not yet, granted there will be some scenarios where new products will be used vs. others, or existing ones used in new ways for different things.
As has been the case for years if not decades, some products will survive, continue to evolve and find new roles, kind of like different data storage mediums (e.g. ssd, disk, tape, etc).
How does ViPR work?
ViPR functions as a control plane across the data and storage infrastructure supporting both north and southbound. northbound refers to use from or up to application servers (physical machines PM and virtual machines VMs). southbound refers target or destination storage systems. Storage systems can be traditional EMC or third-party (NetApp mentioned as part of first release), appliances, just a bunch of disks (JBOD) or cloud services.
Some general features and functions:
Provisioning and allocation (with automation)
Data and storage migration or tiering
Leverage scripts, templates and workbooks
Support service categories and catalogs
Discovery, registration of storage systems
Create of storage resource pools for host systems
Metering, measuring, reporting, charge or show back
Alerts, alarms and notification
Self-service portal for access and provisioning
ViPR data plane (adding data services and value when needed)
Another part is the data plane for implementing data services and access. For block and file when not needed, ViPR steps out-of-the-way leveraging the underlying storage systems or services.
Object storage access
When needed, the ViPR data plane can step in to add added services and functionality along with support object based access for little data and big data. For example, Hadoop Distributed File System (HDFS) services can support northbound analytic software applications running on servers accessing storage managed by ViPR.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved
During the recent EMCworld event in Las Vegas among other things, EMC announced ViPR (read announcement here) . Note that this ViPR is not the same EMC Viper project from a few years ago that was focused on data footprint reduction (DFR) including dedupe. ViPR has been in the works for a couple of years taking a step back rethinking how storage is can be used going forward.
ViPR is not a technology creation developed in a vacuum instead includes customer feedback, wants and needs. Its core themes are extensible, open and scalable.
On the other hand, ViPR addresses plenty of buzzword bingo themes including:
IT as a Service (ITaaS) and Infrastructure as a Service (IaaS)
Scaling with stability without compromise
Software defined storage
Public, private, hybrid cloud
Big data and little data
Block, file and object storage
Control plane and data plane
Storage hypervisor, virtualization and virtual storage
Heterogeneous (third-party) storage support
Open API and automation
Self-service portals, service catalogs
Note that this is essentially announcing the ViPR product and program initiative with general availability slated for second half of 2013.
What is ViPR addressing?
IT and data infrastructure (server, storage, IO and networking hardware, software) challenges for traditional, virtual and cloud environments.
Data growth, after all, there is no such thing as an information recession with more data being generated, moved, processed, stored and retained for longer periods of time. Then again, people and data are both getting larger and living longer, for both little data and big data along with very big data.
Overhead and complexities associated with managing and using an expanding, homogenous (same vendor, perhaps different products) or heterogeneous (different vendors and products) data infrastructure across cloud, virtual and physical, legacy and emerging. This includes add, changes or moves, updates and upgrades, retirement and replacement along with disposition, not to mention protecting data in an expanding footprint.
Operations and service management, fault and alarm notification, resolution and remediation, rapid provisioning, removing complexity and cost of doing things vs. simply cutting cost and compromising service.
What is this software defined storage stuff?
There is the buzzword aspect, and then there is the solution and business opportunity.
First the buzzword aspect and bandwagon:
Software defined marketing (SDM) Leveraging software defined buzzwords.
Software defined data centers (SDDC) Leveraging software to derive more value from hardware while enabling agility, flexibility, and scalability and removing complexity. Think the Cloud and Virtual Data Center models including those from VMware among others.
Software defined networking (SDN) Rather than explain, simply look at Nicira that VMware bought in 2012.
Software defined storage (SDS) Storage software that is independent of any specific hardware, which might be a bit broad, however it is also narrower than saying anything involving software.
Software defined BS (SDBS) Something that usually happens as a result when marketers and others jump on a bandwagon, in this case software defined marketing.
Note that not everything involved with software defined is BS, only some of the marketing spins and overuse. The downside to the software defined marketing and SDBS is the usual reaction of skepticism, cynicism and dismissal, so let us leave the software defined discussion here for now.
An example of software defined storage can be storage virtualization, virtual storage and storage hypervisors that are hardware independent. Note that when I say hardware independent, that also means being able to support different vendors systems. Now if you want to have some fun with the software defined storage diehards or purist, tell them that all hardware needs software and all software needs hardware, even if virtual. Further hardware is defined by its software, however lets leave sleeping dogs lay where they rest (at least for now ;)).
Regardless of what you view of software defined storage, storage hypervisor, storage virtualization and virtual storage is, the primary focus and goal should be addressing business and application needs. Unfortunately, some of the discussions or debates about what is or is not software defined and related themes lose focus of what should be the core goal of enabling business and applications.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved
Cloud Bulk Big Data Software Defined Object Storage Resources
Welcome to the Cloud, Big Data, Software Defined, Bulk and Object Storage Resources Center Page objectstoragecenter.com.
This object storage resources, along with software defined, cloud, bulk, and scale-out storage page is part of the server StorageIOblog microsite collection of resources. Software-defined, Bulk, Cloud and Object Storage exist to support expanding and diverse application data demands.
Bulk, Cloud, Object Storage Solutions and Services
There are various types of cloud, bulk, and object storage including public services such as Amazon Web Services (AWS) Simple Storage Service (S3), Backblaze, Google, Microsoft Azure, IBM Softlayer, Rackspace among many others. There are also solutions for hybrid and private deployment from Cisco, Cloudian, CTERA, Cray, DDN, Dell EMC, Elastifile, Fujitsu, Vantera/HDS, HPE, Hedvig, Huawei, IBM, NetApp, Noobaa, OpenIO, OpenStack, Quantum, Rackspace, Rozo, Scality, Spectra, Storpool, StorageCraft, Suse, Swift, Virtuozzo, WekaIO, WD, among many others.
Cloud products and services among others, along with associated data infrastructures including object storage, file systems, repositories and access methods are at the center of bulk, big data, big bandwidth and little data initiatives on a public, private, hybrid and community basis. After all, not everything is the same in cloud, virtual and traditional data centers or information factories from active data to in-active deep digital archiving.
Object Context Matters
Before discussing Object Storage lets take a step back and look at some context that can clarify some confusion around the term object. The word object has many different meanings and context, both inside of the IT world as well as outside. Context matters with the term object such as a verb being a thing that can be seen or touched as well as a person or thing of action or feeling directed towards.
Besides a person, place or physical thing, an object can be a software-defined data structure that describes something. For example, a database record describing somebody’s contact or banking information, or a file descriptor with name, index ID, date and time stamps, permissions and access control lists along with other attributes or metadata. Another example is an object or blob stored in a cloud or object storage system repository, as well as an item in a hypervisor, operating system, container image or other application.
Besides being a verb, an object can also be a noun such as disapproval or disagreement with something or someone. From an IT context perspective, an object can also refer to a programming method (e.g. object-oriented programming [oop], or Java [among other environments] objects and classes) and systems development in addition to describing entities with data structures.
In other words, a data structure describes an object that can be a simple variable, constant, complex descriptor of something being processed by a program, as well as a function or unit of work. There are also objects unique or with context to specific environments besides Java or databases, operating systems, hypervisors, file systems, cloud and other things.
The Need For Bulk, Cloud and Object Storage
There is no such thing as an information recession with more data being generated, moved, processed, stored, preserved and served, granted there are economic realities. Likewise as a society our dependence on information being available for work or entertainment, from medical healthcare to social media and all points in between continues to increase (check out the Human Face of Big Data).
Object and cloud storage are in your future, the questions are when, where, with what and how among others.
Watch for more content and links to be added here soon to this object storage center page including posts, presentations, pod casts, polls, perspectives along with services and product solutions profiles.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.
As luck or fate would turn out, I was in Nijkerk which is about an hour train ride from Amsterdam central station plus a free day in my schedule. After a morning train ride and nice walk from Amsterdam Central I arrived at the Tobacco Theatre (a former tobacco trading venue) where Ceph Day was underway, and in time for lunch of Krokettens sandwich.
Lets take a quick step back and address for those not familiar what is Ceph (Cephalanthera) and why it was worth spending a day to attend this event. Ceph is an open source distributed object scale out (e.g. cluster or grid) software platform running on industry standard hardware.
Ceph is used for deploying object storage, cloud storage and managed services, general purpose storage for research, commercial, scientific, high performance computing (HPC) or high productivity computing (commercial) along with backup or data protection and archiving destinations. Other software similar in functionality or capabilities to Ceph include OpenStack Swift, Basho Riak CS, Cleversafe, Scality and Caringo among others. There are also the tin wrapped software (e.g. appliances or pre-packaged) solutions such as Dell DX (Caringo), DataDirect Networks (DDN) WOS, EMC ATMOS and Centera, Amplidata and HDS HCP among others. From a service standpoint, these solutions can be used to build services similar Amazon S3 and Glacier, Rackspace Cloud files and Cloud Block, DreamHost DreamObject and HP Cloud storage among others.
At the heart of Ceph is RADOS a distributed object store that consists of peer nodes functioning as object storage devices (OSD). Data can be accessed via REST (Amazon S3 like) APIs, Libraries, CEPHFS and gateway with information being spread across nodes and OSDs using a CRUSH based algorithm (note Sage Weil is one of the authors of CRUSH: Controlled, Scalable, Decentralized Placement of Replicated Data). Ceph is scalable in terms of performance, availability and capacity by adding extra nodes with hard disk drives (HDD) or solid state devices (SSDs). One of the presentations pertained to DreamHost that was an early adopter of Ceph to make their DreamObjects (cloud storage) offering.
In addition to storage nodes, there are also an odd number of monitor nodes to coordinate and manage the Ceph cluster along with optional gateways for file access. In the above figure (via DreamHost), load balancers sit in front of gateways that interact with the storage nodes. The storage node in this example is a physical server with 12 x 3TB HDDs each configured as a OSD.
In the DreamHost example above, there are 90 storage nodes plus 3 management nodes, the total raw storage capacity (no RAID) is about 3PB (12 x 3TB = 36TB x 90 = 3.24PB). Instead of using RAID or mirroring, each objects data is replicated or copied to three (e.g. N=3) different OSDs (on separate nodes), where N is adjustable for a given level of data protection, for a usable storage capacity of about 1PB.
Note that for more usable capacity and lower availability, N could be set lower, or a larger value of N would give more durability or data protection at higher storage capacity overhead cost. In addition to using JBOD configurations with replication, Ceph can also be configured with a combination of RAID and replication providing more flexibility for larger environments to balance performance, availability, capacity and economics.
One of the benefits of Ceph is the flexibility to configure it how you want or need for different applications. This can be in a cost-effective hardware light configuration using JBOD or internal HDDs in small form factor generally available servers, or high density servers and storage enclosures with optional RAID adapters along with SSD. This flexibility is different from some cloud and object storage systems or software tools which take a stance of not using or avoiding RAID vs. providing options and flexibility to configure and use the technology how you see fit.
Here are some links to presentations from Ceph Day: Introduction and Welcome by Wido den Hollander Ceph: A Unified Distributed Storage System by Sage Weil Ceph in the Cloud by Wido den Hollander DreamObjects: Cloud Object Storage with Ceph by Ross Turk Cluster Design and Deployment by Greg Farnum Notes on Librados by Sage Weil
While at Ceph day, I was able to spend a few minutes with Sage Weil Ceph creator and founder of inktank.com to record a pod cast (listen here) about what Ceph is, where and when to use it, along with other related topics. Also while at the event I had a chance to sit down with Curtis (aka Mr. Backup) Preston where we did a simulcast video and pod cast. The simulcast involved Curtis recording this video with me as a guest discussing Ceph, cloud and object storage, backup, data protection and related themes while I recorded this pod cast.
One of the interesting things I heard, or actually did not hear while at the Ceph Day event that I tend to hear at related conferences such as SNW is a focus on where and how to use, configure and deploy Ceph along with various configuration options, replication or copy modes as opposed to going off on erasure codes or other tangents. In other words, instead of focusing on the data protection protocol and algorithms, or what is wrong with the competition or other architectures, the Ceph Day focused was removing cloud and object storage objections and enablement.
Where do you get Ceph? You can get it here, as well as via 42on.com and inktank.com.
Trick or treat and vendor fun games Updated 6/26/18
In the spirit of Halloween and zombies season, a couple of thoughts come to mind about vendor tricks and treats. This is an industry trends and perspectives post, part of an ongoing series looking at various technology and fun topics.
The first trick or treat game pertains to the blame game; you know either when something breaks, or at the other extreme, before you have even made a decision to buy something. The trick or treat game for decision-making goes something like this.
Vendor “A” says products succeed with their solution while failure results with a solution from “B” when doing “X”. Otoh, vendor “B” claims that “X” will fail when using a solution from vendor “A”. In fact, you can pick what you want to substitute for “X”, perhaps VDI, Big Data, Little Data, Backup, Archive, Analytics, Private Cloud, Public Cloud, Hybrid Cloud, eDiscovery you name it.
This is not complicated math or big data problem requiring a high-performance computing (HPC) platform. A HPC Zetta-Flop processing ability using 512 bit addressing of 9.9 (e.g. 1 nine) PettaBytes of battery-backed DRAM and an IO capability of 9.99999 (e.g. 5 9’s) trillion 8 bit IOPS to do table pivots or runge kutta numerical analysis, map reduce, SAS or another modeling with optional iProduct or Android interface are not needed.
StorageIO images of touring Texas Advanced Computing (e.g. HPC) Center
Can you solve this equation? Hint it does not need a PhD or any other advanced degree. Another hint, if you have ever been at any side of the technology product and services decision-making table, regardless of the costume you wore, you should know the answer.
Of course the question of would “X” fail regardless of who or what “A” or “B” let alone a “C”, “D” or “F”? In other words, it is not the solution, technology, vendor or provider, rather the problem or perhaps even lack thereof that is the issue. Or is it a case where there is a solution from “A”, “B” or any others that is looking for a problem, and if it is the wrong problem, there can be a wrong solution thus failure?
Another trick or treat game is vendors public relations (PR) or analyst relations (AR) people to ask for one thing and delivery or ask another. For example, some vendor, service provider, their marketing AR and PR people or surrogates make contact wanting to tell of various success and failure story. Of course, this is usually their success and somebody else’s failure, or their victory over something or someone who sometimes can be interesting. Of course, there are also the treats to get you to listen to the above, such as tempt you with a project if you meet with their subject, which may be a trick of a disappearing treat (e.g. magic, poof it is gone after the discussion).
There are another AR and PR trick and treat where they offer on behalf of their representative organization or client to a perspective or exclusive insight on their competitor. Of course, the treat from their perspective is that they will generously expose all that is wrong with what a competitor is saying about their own (e.g. the competitors) product.
Let me get this straight, I am not supposed to believe what somebody says about his or her own product, however, supposed to believe what a competitor says is wrong with the competition’s product, and what is right with his or her own product.
Hmm, ok, so let me get this straight, a competitor say “A” wants to tell me what somebody say from “B” has told me is wrong and I should schedule a visit with a truth squad member from “A” to get the record set straight about “B”?
Does that mean then that I go to “B” for a rebuttal, as well as an update about “A” from “B”, assuming that what “A” has told me is also false about themselves, and perhaps about “B” or any other?
Too be fair, depending on your level of trust and confidence in either a vendor, their personal or surrogates, you might tend to believe more from them vs. others, or at least until you been tricked after given treats. There may be some that have been tricked, or they tried applying to many treats to present a story that behind the costume might be a bit scary.
Having been through enough of these, and I candidly believe that sometimes “A” or “B” or any other party actually do believe that they have more or better info about their competitor and that they can convince somebody about what their competitor is doing better than the competitor can. I also believe that there are people out there who will go to “A” or “B” and believe what they are told by based on their preference, bias or interests.
When I hear from vendors, VARs, solution or service providers and others, it’s interesting hearing point, counterpoint and so forth, however if time is limited, I’am more interested in hearing from such as “A” about them, what they are doing, where success, where challenges, where going and if applicable, under NDA go into more detail.
Customer success stories are good, however again, if interested in what works, what kind of works, or what does not work, chances are when looking for G2 vs. GQ, a non-scripted customer conversation or perspective of the good, the bad and the ugly is preferred, even if under NDA. Again, if time is limited which it usually is, focus on what is being done with your solution, where it is going and if compelled send follow-up material that can of course include MUD and FUD about others if that is your preference.
Then there is when during a 30 minute briefing, the vendor or solution provider is still talking about trends, customer pain points, what competitors are doing at 21 minutes into the call with no sign of an announcement, update or news in site
Lets not forget about the trick where the vendor marketing or PR person reaches out and says that the CEO, CMO, CTO or some other CxO or Chief Jailable Officer (CJO) wants to talk with you. Part of the trick is when the CxO actually makes it to the briefing and is not ready, does not know why the call is occurring, or, thinks that a request for an audience has been made with them for an interview or something else.
A treat is when 3 to 4 minutes into a briefing, the vendor or solution provider has already framed up what and why they are doing something. This means getting to what they are announcing or planning on doing and getting into a conversation to discuss what they are doing and making good follow-up content and resources available.
Sometimes a treat is when a briefer goes on autopilot nailing their script for 29 of a 30 minute session then use the last-minute to ask if there are any questions. The reason autopilot briefings can be a treat is when they are going over what is in the slide deck, webex, or press release thus affording an opportunity to get caught up on other things while talk at you. Hmm, perhaps need to consider playing some tricks in reward for those kind of treats? ;)
Do not be scared, not everybody is out to trick you with treats, and not all treats have tricks attached to them. Be prepared, figure out who is playing tricks with treats, and who has treats without tricks.
Oh, and as a former IT customer, vendor and analyst, one of my favorites is contact information of my dogs to vendors who require registration on their websites for basic things such as data sheets. Another is supplying contact information of competing vendors sales reps to vendors who also require registration for basic data sheets or what should otherwise be generally available information as opposed to more premium treats. Of course there are many more fun tricks, however lets leave those alone for now.
Note: Zombie voting rules apply which means vote early, vote often, and of course vote for those who cannot include those that are dead (real or virtual).
Where To Learn More
View additiona related material via the following links.
All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.
Before announcing Glacier, options for Amazon storage services relied on general purpose S3, or EBS with other Amazon services. S3 has provided users the ability to select different availability zones (e.g. geographical regions where data is stored) along with level of reliability for different price points for their applications or services being offered.
Note that AWS S3 flexibility lends itself to individuals or organizations using it for various purposes. This ranges from storing backup or file sharing data to being used as a target for other cloud services. S3 pricing options vary depending on which availability zones you select as well as if standard or reduced redundancy. As its name implies, reduced redundancy trades lower availability recovery time objective (RTO) in exchange for lower cost per given amount of space capacity.
AWS has now announced a new class or tier of storage service called Glacier, which as its name implies moves very slow and capable of supporting large amounts of data. In other words, targeting inactive or seldom accessed data where emphasis is on ultra-low cost in exchange for a longer RTO. In exchange for an RTO that AWS is stating that it can be measured in hours, your monthly storage cost can be as low as 1 cent per GByte or about 12 cents per year per GByte plus any extra fees (See here).
Here is a note that I received from the Amazon Web Services (AWS) team:
Dear Amazon Web Services Customer, We are excited to announce the immediate availability of Amazon Glacier – a secure, reliable and extremely low cost storage service designed for data archiving and backup. Amazon Glacier is designed for data that is infrequently accessed, yet still important to keep for future reference. Examples include digital media archives, financial and healthcare records, raw genomic sequence data, long-term database backups, and data that must be retained for regulatory compliance. With Amazon Glacier, customers can reliably and durably store large or small amounts of data for as little as $0.01/GB/month. As with all Amazon Web Services, you pay only for what you use, and there are no up-front expenses or long-term commitments.
Low cost– Amazon Glacier is an extremely low-cost, pay-as-you-go storage service that can cost as little as $0.01 per gigabyte per month, irrespective of how much data you store.
Secure – Amazon Glacier supports secure transfer of your data over Secure Sockets Layer (SSL) and automatically stores data encrypted at rest using Advanced Encryption Standard (AES) 256, a secure symmetrix-key encryption standard using 256-bit encryption keys.
Durable– Amazon Glacier is designed to give average annual durability of 99.999999999% for each item stored.
Flexible -Amazon Glacier scales to meet your growing and often unpredictable storage requirements. There is no limit to the amount of data you can store in the service.
Simple– Amazon Glacier allows you to offload the administrative burdens of operating and scaling archival storage to AWS, and makes long term data archiving especially simple. You no longer need to worry about capacity planning, hardware provisioning, data replication, hardware failure detection and repair, or time-consuming hardware migrations.
Designed for use with other Amazon Web Services – You can use AWS Import/Export to accelerate moving large amounts of data into Amazon Glacier using portable storage devices for transport. In the coming months, Amazon Simple Storage Service (Amazon S3) plans to introduce an option that will allow you to seamlessly move data between Amazon S3 and Amazon Glacier using data lifecycle policies.
Amazon Glacier is currently available in the US-East (N. Virginia), US-West (N. California), US-West (Oregon), EU-West (Ireland), and Asia Pacific (Japan) Regions.
Glacier is low-cost for lower performance (e.g. access time) storage suited to data applications including archiving, inactive or idle data that you are not in a hurry to retrieve. Pay as you go pricing that can be as low as $0.01 USD per GByte per month (and other optional fees may apply, see here) depending on availability zone. Availability zone or regions include US West coast (Oregon or Northern California), US East Coast (Northern Virginia), Europe (Ireland) and Asia (Tokyo).
Now what is understood should have to be discussed, however just to be safe, pity the fool who complains about signing up for AWS Glacier due to its penny per month per GByte cost and it being too slow for their iTunes or videos as you know its going to happen. Likewise, you know that some creative vendor or their surrogate is going to try to show a miss-match of AWS Glacier vs. their faster service that caters to a different usage model; it is just a matter of time.
Lets be clear, Glacier is designed for low-cost, high-capacity, slow access of infrequently accessed data such as an archive or other items. This means that you will be more than disappointed if you try to stream a video, or access a document or photo from Glacier as you would from S3 or EBS or any other cloud service. The reason being is that Glacier is designed with the premise of low-cost, high-capacity, high availability at the cost of slow access time or performance. How slow? AWS states that you may have to wait several hours to reach your data when needed, however that is the tradeoff. If you need faster access, pay more or find a different class and tier of storage service to meet that need, perhaps for those with the real need for speed, AWS SSD capabilities ;).
Here is a link to a good post over at Planforcloud.com comparing Glacier vs. S3, which is like comparing apples and oranges; however, it helps to put things into context.
In terms of functionality, Glacier security includes secure socket layer (SSL), advanced encryption standard (AES) 256 (256-bit encryption keys) data at rest encryption along with AWS identify and access management (IAM) policies.
Persistent storage designed for 99.999999999% durability with data automatically placed in different facilities on multiple devices for redundancy when data is ingested or uploaded. Self-healing is accomplished with automatic background data integrity checks and repair.
Scale and flexibility are bound by the size of your budget or credit card spending limit along with what availability zones and other options you choose. Integration with other AWS services including Import/Export where you can ship large amounts of data to Amazon using different media and mediums. Note that AWS has also made a statement of direction (SOD) that S3 will be enhanced to seamless move data in and out of Glacier using data policies.
Part of stretching budgets for organizations of all size is to avoid treating all data and applications the same (key theme of data protection modernization). This means classifying and addressing how and where different applications and data are placed on various types of servers, storage along with revisiting modernizing data protection.
While the low-cost of Amazon Glacier is an attention getter, I am looking for more than just the lowest cost, which means I am also looking for reliability, security among other things to gain and keep confidence in my cloud storage services providers. As an example, a few years ago I switched from one cloud backup provider to another not based on cost, rather functionality and ability to leverage the service more extensively. In fact, I could switch back to the other provider and save money on the monthly bills; however I would end up paying more in lost time, productivity and other costs.
What do I see as the barrier to AWS Glacier adoption?
Simple, getting vendors and other service providers to enhance their products or services to leverage the new AWS Glacier storage category. This means backup/restore, BC and DR vendors ranging from Amazon (e.g. releasing S3 to Glacier automated policy based migration), Commvault, Dell (via their acquisitions of Appassure and Quest), EMC (Avamar, Networker and other tools), HP, IBM/Tivoli, Jungledisk/Rackspace, NetApp, Symantec and others, not to mention cloud gateway providers will need to add support for this new capabilities, along with those from other providers.
As an Amazon EC2 and S3 customer, it is great to see Amazon continue to expand their cloud compute, storage, networking and application service offerings. I look forward to actually trying out Amazon Glacier for storing encrypted archive or inactive data to compliment what I am doing. Since I am not using the Amazon Cloud Storage Gateway, I am looking into how I can use Rackspace Jungledisk to manage an Amazon Glacier repository similar to how it manages my S3 stores.
As of now, it looks like I will have to wait for either Jungledisk adds native support as they do today for managing my S3 storage pool today, or, the automated policy based movement between S3 and Glacier is transparently enabled.
