Tag: tpc

Posted on

Which Enterprise HDD for Content Server Platform

Which Enterprise HDD to use for a Content Server Platform

data infrastructure HDD server storage I/O trends

Updated 1/23/2018

Which enterprise HDD to use with a content server platform?

Insight for effective server storage I/O decision making
Server StorageIO Lab Review

Which enterprise HDD to use for content servers

This post is the first in a multi-part series based on a white paper hands-on lab report I did compliments of Equus Computer Systems and Seagate that you can read in PDF form here. The focus is looking at the Equus Computer Systems (www.equuscs.com) converged Content Solution platforms with Seagate Enterprise Hard Disk Drive (HDD’s). I was given the opportunity to do some hands-on testing running different application workloads with a 2U content solution platform along with various Seagate Enterprise 2.5” HDD’s handle different application workloads. This includes Seagate’s Enterprise Performance HDD’s with the enhanced caching feature.

Issues And Challenges

Even though Non-Volatile Memory (NVM) including NAND flash solid state devices (SSDs) have become popular storage for use internal as well as external to servers, there remains the need for HDD’s Like many of you who need to make informed server, storage, I/O hardware, software and configuration selection decisions, time is often in short supply.

A common industry trend is to use SSD and HDD based storage mediums together in hybrid configurations. Another industry trend is that HDD’s continue to be enhanced with larger space capacity in the same or smaller footprint, as well as with performance improvements. Thus, a common challenge is what type of HDD to use for various content and application workloads balancing performance, availability, capacity and economics.

Content Applications and Servers

Fast Content Needs Fast Solutions

An industry and customer trend are that information and data are getting larger, living longer, as well as there is more of it. This ties to the fundamental theme that applications and their underlying hardware platforms exist to process, move, protect, preserve and serve information.

Content solutions span from video (4K, HD, SD and legacy streaming video, pre-/post-production, and editing), audio, imaging (photo, seismic, energy, healthcare, etc.) to security surveillance (including Intelligent Video Surveillance [ISV] as well as Intelligence Surveillance and Reconnaissance [ISR]). In addition to big fast data, other content solution applications include content distribution network (CDN) and caching, network function virtualization (NFV) and software-defined network (SDN), to cloud and other rich unstructured big fast media data, analytics along with little data (e.g. SQL and NoSQL database, key-value stores, repositories and meta-data) among others.

Content Solutions And HDD Opportunities

A common theme with content solutions is that they get defined with some amount of hardware (compute, memory and storage, I/O networking connectivity) as well as some type of content software. Fast content applications need fast software, multi-core processors (compute), large memory (DRAM, NAND flash, SSD and HDD’s) along with fast server storage I/O network connectivity. Content-based applications benefit from having frequently accessed data as close as possible to the application (e.g. locality of reference).

Content solution and application servers need flexibility regarding compute options (number of sockets, cores, threads), main memory (DRAM DIMMs), PCIe expansion slots, storage slots and other connectivity. An industry trend is leveraging platforms with multi-socket processors, dozens of cores and threads (e.g. logical processors) to support parallel or high-concurrent content applications. These servers have large amounts of local storage space capacity (NAND flash SSD and HDD) and associated I/O performance (PCIe, NVMe, 40 GbE, 10 GbE, 12 Gbps SAS etc.) in addition to using external shared storage (local and cloud).

Where To Learn More

Additional learning experiences along with common questions (and answers), as well as tips can be found in Software Defined Data Infrastructure Essentials book.

Software Defined Data Infrastructure Essentials Book SDDC

What This All Means

Fast content applications need fast content and flexible content solution platforms such as those from Equus Computer Systems and HDD’s from Seagate. Key to a successful content application deployment is having the flexibility to hardware define and software defined the platform to meet your needs. Just as there are many different types of content applications along with diverse environments, content solution platforms need to be flexible, scalable and robust, not to mention cost effective.

Continue reading part two of this multi-part series here where we look at how and what to test as well as project planning.

Ok, nuff said, for now.

Gs

Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

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.

Posted on

Part 4 – Which HDD for Content Applications – Database Workloads

Part 4 – Which HDD for Content Applications – Database Workloads

data base server storage I/O trends

Updated 1/23/2018
Which enterprise HDD to use with a content server platform for database workloads

Insight for effective server storage I/O decision making
Server StorageIO Lab Review

Which enterprise HDD to use for content servers

This is the fourth in a multi-part series (read part three here) based on a white paper hands-on lab report I did compliments of Servers Direct and Seagate that you can read in PDF form here. The focus is looking at the Servers Direct (www.serversdirect.com) converged Content Solution platforms with Seagate Enterprise Hard Disk Drive (HDD’s). In this post the focus expands to database application workloads that were run to test various HDD’s.

Database Reads/Writes

Transaction Processing Council (TPC) TPC-C like workloads were run against the SUT from the STI. These workloads simulated transactional, content management, meta-data and key-value processing. Microsoft SQL Server 2012 was configured and used with databases (each 470GB e.g. scale 6000) created and workload generated by virtual users via Dell Benchmark Factory (running on STI Windows 2012 R2).

A single SQL Server database instance (8) was used on the SUT, however unique databases were created for each HDD set being tested. Both the main database file (.mdf) and the log file (.ldf) were placed on the same drive set being tested, keep in mind the constraints mentioned above. As time was a constraint, database workloads were run concurrent (9) with each other except for the Enterprise 10K RAID 1 and RAID 10. Workload was run with two 10K HDD’s in a RAID 1 configuration, then another workload run with a four drive RAID 10. In a production environment, ideally the .mdf and .ldf would be placed on separate HDD’s and SSDs.

To improve cache buffering the SQL Server database instance memory could be increased from 16GB to a larger number that would yield higher TPS numbers. Keep in mind the objective was not to see how fast I could make the databases run, rather how the different drives handled the workload.

(Note 8) The SQL Server Tempdb was placed on a separate NVMe flash SSD, also the database instance memory size was set to 16GB which was shared by all databases and virtual users accessing it.

(Note 9) Each user step was run for 90 minutes with a 30 minute warm-up preamble to measure steady-state operation.

Users

TPCC Like TPS

Single Drive Cost per TPS

Drive Cost per TPS

Single Drive Cost / Per GB Raw Cap.

Cost / Per GB Usable (Protected) Cap.

Drive Cost (Multiple Drives)

Protect
Space Over head

Cost per usable GB per TPS

Resp. Time (Sec.)

ENT 15K R1

1

23.9

$24.94

$49.89

$0.99

$0.99

$1,190

100%

$49.89

0.01

ENT 10K R1

1

23.4

$37.38

$74.77

$0.49

$0.49

$1,750

100%

$74.77

0.01

ENT CAP R1

1

16.4

$24.26

$48.52

$0.20

$0.20

$ 798

100%

$48.52

0.03

ENT 10K R10

1

23.2

$37.70

$150.78

$0.49

$0.97

$3,500

100%

$150.78

0.07

ENT CAP SWR5

1

17.0

$23.45

$117.24

$0.20

$0.25

$1,995

20%

$117.24

0.02

ENT 15K R1

20

362.3

$1.64

$3.28

$0.99

$0.99

$1,190

100%

$3.28

0.02

ENT 10K R1

20

339.3

$2.58

$5.16

$0.49

$0.49

$1,750

100%

$5.16

0.01

ENT CAP R1

20

213.4

$1.87

$3.74

$0.20

$0.20

$ 798

100%

$3.74

0.06

ENT 10K R10

20

389.0

$2.25

$9.00

$0.49

$0.97

$3,500

100%

$9.00

0.02

ENT CAP SWR5

20

216.8

$1.84

$9.20

$0.20

$0.25

$1,995

20%

$9.20

0.06

ENT 15K R1

50

417.3

$1.43

$2.85

$0.99

$0.99

$1,190

100%

$2.85

0.08

ENT 10K R1

50

385.8

$2.27

$4.54

$0.49

$0.49

$1,750

100%

$4.54

0.09

ENT CAP R1

50

103.5

$3.85

$7.71

$0.20

$0.20

$ 798

100%

$7.71

0.45

ENT 10K R10

50

778.3

$1.12

$4.50

$0.49

$0.97

$3,500

100%

$4.50

0.03

ENT CAP SWR5

50

109.3

$3.65

$18.26

$0.20

$0.25

$1,995

20%

$18.26

0.42

ENT 15K R1

100

190.7

$3.12

$6.24

$0.99

$0.99

$1,190

100%

$6.24

0.49

ENT 10K R1

100

175.9

$4.98

$9.95

$0.49

$0.49

$1,750

100%

$9.95

0.53

ENT CAP R1

100

59.1

$6.76

$13.51

$0.20

$0.20

$ 798

100%

$13.51

1.66

ENT 10K R10

100

560.6

$1.56

$6.24

$0.49

$0.97

$3,500

100%

$6.24

0.14

ENT CAP SWR5

100

62.2

$6.42

$32.10

$0.20

$0.25

$1,995

20%

$32.10

1.57

Table-2 TPC-C workload results various number of users across different drive configurations

Figure-2 shows TPC-C TPS (red dashed line) workload scaling over various number of users (1, 20, 50, and 100) with peak TPS per drive shown. Also shown is the used space capacity (in green), with total raw storage capacity in blue cross hatch. Looking at the multiple metrics in context shows that the 600GB Enterprise 15K HDD with performance enhanced cache is a premium option as an alternative, or, to complement flash SSD solutions.

database TPCC transactional workloads
Figure-2 472GB Database TPS scaling along with cost per TPS and storage space used

In figure-2, the 1.8TB Enterprise 10K HDD with performance enhanced cache while not as fast as the 15K, provides a good balance of performance, space capacity and cost effectiveness. A good use for the 10K drives is where some amount of performance is needed as well as a large amount of storage space for less frequently accessed content.

A low cost, low performance option would be the 2TB Enterprise Capacity HDD’s that have a good cost per capacity, however lack the performance of the 15K and 10K drives with enhanced performance cache. A four drive RAID 10 along with a five drive software volume (Microsoft WIndows) are also shown. For apples to apples comparison look at costs vs. capacity including number of drives needed for a given level of performance.

Figure-3 is a variation of figure-2 showing TPC-C TPS (blue bar) and response time (red-dashed line) scaling across 1, 20, 50 and 100 users. Once again the Enterprise 15K with enhanced performance cache feature enabled has good performance in an apples to apples RAID 1 comparison.

Note that the best performance was with the four drive RAID 10 using 10K HDD’s Given popularity, a four drive RAID 10 configuration with the 10K drives was used. Not surprising the four 10K drives performed better than the RAID 1 15Ks. Also note using five drives in a software spanned volume provides a large amount of storage capacity and good performance however with a larger drive footprint.

database TPCC transactional workloads scaling
Figure-3 472GB Database TPS scaling along with response time (latency)

From a cost per space capacity perspective, the Enterprise Capacity drives have a good cost per GB. A hybrid solution for environment that do not need ultra-high performance would be to pair a small amount of flash SSD (10) (drives or PCIe cards), as well as the 10K and 15K performance enhanced drives with the Enterprise Capacity HDD (11) along with cache or tiering software.

(Note 10) Refer to Seagate 1200 12 Gbps Enterprise SAS SSD StorageIO lab review

(Note 11) Refer to Enterprise SSHD and Flash SSD Part of an Enterprise Tiered Storage Strategy

Where To Learn More

Additional learning experiences along with common questions (and answers), as well as tips can be found in Software Defined Data Infrastructure Essentials book.

Software Defined Data Infrastructure Essentials Book SDDC

What This All Means

If your environment is using applications that rely on databases, then test resources such as servers, storage, devices using tools that represent your environment. This means moving up the software and technology stack from basic storage I/O benchmark or workload generator tools such as Iometer among others instead using either your own application, or tools that can replay or generate various workloads that represent your environment.

Continue reading part five in this multi-part series here where the focus shifts to large and small file I/O processing workloads.

Ok, nuff said, for now.

Gs

Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

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.

Posted on

Collecting Transaction Per Minute from SQL Server and HammerDB

Storage I/O trends

Collecting Transaction Per Minute from SQL Server and HammerDB

When using benchmark or workload generation tools such as HammerDB I needed a way to capture and log performance activity metrics such as transactions per minute. For example using HammerDB to simulate an application making database requests performing various transactions as part of testing an overall system solution including server and storage I/O activity. This post takes a look at the problem or challenge I was looking to address, as well as creating a solution after spending time searching for one (still searching btw).

The Problem, Issue, Challenge, Opportunity and Need

The challenge is to collect application performance such as transactions per minute from a workload using a database. The workload or benchmark tool (in this case HammerDB) is the System Test Initiator (STI) that drives the activity (e.g. database requests) to a System Under Test (SUT). In this example the SUT is a Microsoft SQL Server running on a Windows 2012 R2 server. What I need is to collect and log into a file for later analysis the transaction rate per minute while the STI is generating a particular workload.

Server Storage I/O performance

Understanding the challenge and designing a strategy

If you have ever used benchmark or workload generation tools such as Quest Benchmark Factory (part of the Toad tools collection) you might be spoiled with how it can be used to not only generate the workload, as well as collect, process, present and even store the results for database workloads such as TPC simulations. In this situation, Transaction Processing Council (TPC) like workloads need to be run and metrics on performance collected. Lets leave Benchmark Factory for a future discussion and focus instead on a free tool called HammerDB and more specifically how to collection transactions per minute metrics from Microsoft SQL Server. While the focus is SQL Server, you can easily adapt the approach for MySQL among others, not to mention there are tools such as Sysbench, Aerospike among other tools.

The following image (created using my Livescribe Echo digital pen) outlines the problem, as well as sketches out a possible solution design. In the following figure, for my solution I’m going to show how to grab every minute for a given amount of time the count of transactions that have occurred. Later in the post processing (you could also do in the SQL Script) I take the new transaction count (which is cumulative) and subtract the earlier interval which yields the transactions per minute (see examples later in this post).

collect TPM metrics from SQL Server with hammerdb
The problem and challenge, a way to collect Transactions Per Minute (TPM)

Finding a solution

HammerDB displays results via its GUI, and perhaps there is a way or some trick to get it to log results to a file or some other means, however after searching the web, found that it was quicker to come up with solution. That solution was to decide how to collect and report the transactions per minute (or you could do by second or other interval) from Microsoft SQL Server. The solution was to find what performance counters and metrics are available from SQL Server, how to collect those and log them to a file for processing. What this means is a SQL Server script file would need to be created that ran in a loop collecting for a given amount of time at a specified interval. For example once a minute for several hours.

Taking action

The following is a script that I came up with that is far from optimal however it gets the job done and is a starting point for adding more capabilities or optimizations.

In the following example, set loopcount to some number of minutes to collect samples for. Note however that if you are running a workload test for eight (8) hours with a 30 minute ramp-up time, you would want to use a loopcount (e.g. number of minutes to collect for) of 480 + 30 + 10. The extra 10 minutes is to allow for some samples before the ramp and start of workload, as well as to give a pronounced end of test number of samples. Add or subtract however many minutes to collect for as needed, however keep this in mind, better to collect a few extra minutes vs. not have them and wished you did.

-- Note and disclaimer:
-- 
-- Use of this code sample is at your own risk with Server StorageIO and UnlimitedIO LLC
-- assuming no responsibility for its use or consequences. You are free to use this as is
-- for non-commercial scenarios with no warranty implied. However feel free to enhance and
-- share those enhancements with others e.g. pay it forward.
-- 
DECLARE @cntr_value bigint;
DECLARE @loopcount bigint; # how many minutes to take samples for

set @loopcount = 240

SELECT @cntr_value = cntr_value
 FROM sys.dm_os_performance_counters
 WHERE counter_name = 'transactions/sec'
 AND object_name = 'MSSQL$DBIO:Databases'
 AND instance_name = 'tpcc' ; print @cntr_value;
 WAITFOR DELAY '00:00:01'
-- 
-- Start loop to collect TPM every minute
-- 

while @loopcount <> 0
begin
SELECT @cntr_value = cntr_value
 FROM sys.dm_os_performance_counters
 WHERE counter_name = 'transactions/sec'
 AND object_name = 'MSSQL$DBIO:Databases'
 AND instance_name = 'tpcc' ; print @cntr_value;
 WAITFOR DELAY '00:01:00'
 set @loopcount = @loopcount - 1
end
-- 
-- All done with loop, write out the last value
-- 
SELECT @cntr_value = cntr_value
 FROM sys.dm_os_performance_counters
 WHERE counter_name = 'transactions/sec'
 AND object_name = 'MSSQL$DBIO:Databases'
 AND instance_name = 'tpcc' ; print @cntr_value;
-- 
-- End of script
--

The above example has loopcount set to 240 for a 200 minute test with a 30 minute ramp and 10 extra minutes of samples. I use the a couple of the minutes to make sure that the system test initiator (STI) such as HammerDB is configured and ready to start executing transactions. You could also put this along with your HammerDB items into a script file for further automation, however I will leave that exercise up to you.

For those of you familiar with SQL and SQL Server you probably already see some things to improve or stylized or simply apply your own preference which is great, go for it. Also note that I’m only selecting a certain variable from the performance counters as there are many others which you can easily discovery with a couple of SQL commands (e.g. select and specify database instance and object name. Also note that the key is accessing the items in sys.dm_os_performance_counters of your SQL Server database instance.

The results

The output from the above is a list of cumulative numbers as shown below which you will need to post process (or add a calculation to the above script). Note that part of running the script is specifying an output file which I show later.

785
785
785
785
37142
1259026
2453479
3635138

Implementing the solution

You can setup the above script to run as part of a larger automation shell or batch script, however for simplicity I’m showing it here using Microsoft SQL Server Studio.

SQL Server script to collect TPM
Microsoft SQL Server Studio with script to collect Transaction Per Minute (TPM)

The following image shows how to specify an output file for the results to be logged to when using Microsoft SQL Studio to run the TPM collection script.

Specify SQL Server tpm output file
Microsoft SQL Server Studio specify output file

With the SQL Server script running to collect results, and HammerDB workload running to generate activity, the following shows Quest Spotlight on Windows (SoW) displaying WIndows Server 2012 R2 operating system level performance including CPU, memory, paging and other activity. Note that this example had about the system test initiator (STI) which is HammerDB and the system under test (SUT) that is Microsoft SQL Server on the same server.

Spotlight on Windows while SQL Server doing tpc
Quest Spotlight on Windows showing Windows Server performance activity

Results and post-processing

As part of post processing simple use your favorite tool or script or what I often do is pull the numbers into Excel spreadsheet, and simply create a new column of numbers that computes and shows the difference between each step (see below). While in Excel then I plot the numbers as needed which can also be done via a shell script and other plotting tools such as R.

In the following example, the results are imported into Excel (your favorite tool or script) where I then add a column (B) that simple computes the difference between the existing and earlier counter. For example in cell B2 = A2-A1, B3 = A3-A2 and so forth for the rest of the numbers in column A. I then plot the numbers in column B to show the transaction rates over time that can then be used for various things.

Hammerdb TPM results from SQL Server processed in Excel
Results processed in Excel and plotted

Note that in the above results that might seem too good to be true they are, these were cached results to show the tools and data collection process as opposed to the real work being done, at least for now…

Where to learn more

Here are some extra links to have a look at:

How to test your HDD, SSD or all flash array (AFA) storage fundamentals
Server and Storage I/O Benchmarking 101 for Smarties
Server and Storage I/O Benchmark Tools: Microsoft Diskspd (Part I)
The SSD Place (collection of flash and SSD resources)
Server and Storage I/O Benchmarking and Performance Resources
I/O, I/O how well do you know about good or bad server and storage I/Os?

What this all means and wrap-up

There are probably many ways to fine tune and optimize the above script, likewise there may even be some existing tool, plug-in, add-on module, or configuration setting that allows HammerDB to log the transaction activity rates to a file vs. simply showing on a screen. However for now, this is a work around that I have found for when needing to collect transaction activity performance data with HammerDB and SQL Server.

Ok, nuff said, for now…

Cheers gs

Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press) and Resilient Storage Networks (Elsevier)
twitter @storageio

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved

Posted on

Server Storage I/O Benchmark Performance Resource Tools

Server Storage I/O Benchmarking Performance Resource Tools

server storage I/O trends

Updated 1/23/2018

Server storage I/O benchmark performance resource tools, various articles and tips. These include tools for legacy, virtual, cloud and software defined environments.

benchmark performance resource tools server storage I/O performance

The best server and storage I/O (input/output operation) is the one that you do not have to do, the second best is the one with the least impact.

server storage I/O locality of reference

This is where the idea of locality of reference (e.g. how close is the data to where your application is running) comes into play which is implemented via tiered memory, storage and caching shown in the figure above.

Cloud virtual software defined storage I/O

Server storage I/O performance applies to cloud, virtual, software defined and legacy environments

What this has to do with server storage I/O (and networking) performance benchmarking is keeping the idea of locality of reference, context and the application workload in perspective regardless of if cloud, virtual, software defined or legacy physical environments.

StorageIOblog: I/O, I/O how well do you know about good or bad server and storage I/Os?
StorageIOblog: Server and Storage I/O benchmarking 101 for smarties
StorageIOblog: Which Enterprise HDDs to use for a Content Server Platform (7 part series with using benchmark tools)
StorageIO.com: Enmotus FuzeDrive MicroTiering lab test using various tools
StorageIOblog: Some server storage I/O benchmark tools, workload scripts and examples (Part I) and (Part II)
StorageIOblog: Get in the NVMe SSD game (if you are not already)
Doridmen.com: Transcend SSD360S Review with tips on using ATTO and Crystal benchmark tools
ComputerWeekly: Storage performance metrics: How suppliers spin performance specifications

Via StorageIO Podcast: Kevin Closson discusses SLOB Server CPU I/O Database Performance benchmarks
Via @KevinClosson: SLOB Use Cases By Industry Vendors. Learn SLOB, Speak The Experts’ Language
Via BeyondTheBlocks (Reduxio): 8 Useful Tools for Storage I/O Benchmarking
Via CCSIObench: Cold-cache Sequential I/O Benchmark
Doridmen.com: Transcend SSD360S Review with tips on using ATTO and Crystal benchmark tools
CISJournal: Benchmarking the Performance of Microsoft Hyper-V server, VMware ESXi and Xen Hypervisors (PDF)
Microsoft TechNet:Windows Server 2016 Hyper-V large-scale VM performance for in-memory transaction processing
InfoStor: What’s The Best Storage Benchmark?
StorageIOblog: How to test your HDD, SSD or all flash array (AFA) storage fundamentals
Via ATTO: Atto V3.05 free storage test tool available
Via StorageIOblog: Big Files and Lots of Little File Processing and Benchmarking with Vdbench

Via StorageIO.com: Which Enterprise Hard Disk Drives (HDDs) to use with a Content Server Platform (White Paper)
Via VMware Blogs: A Free Storage Performance Testing Tool For Hyperconverged
Microsoft Technet: Test Storage Spaces Performance Using Synthetic Workloads in Windows Server
Microsoft Technet: Microsoft Windows Server Storage Spaces – Designing for Performance
BizTech: 4 Ways to Performance-Test Your New HDD or SSD
EnterpriseStorageForum: Data Storage Benchmarking Guide
StorageSearch.com: How fast can your SSD run backwards?
OpenStack: How to calculate IOPS for Cinder Storage ?
StorageAcceleration: Tips for Measuring Your Storage Acceleration

server storage I/O STI and SUT

Spiceworks: Determining HDD SSD SSHD IOP Performance
Spiceworks: Calculating IOPS from Perfmon data
Spiceworks: profiling IOPs

vdbench server storage I/O benchmark
Vdbench example via StorageIOblog.com

StorageIOblog: What does server storage I/O scaling mean to you?
StorageIOblog: What is the best kind of IO? The one you do not have to do
Testmyworkload.com: Collect and report various OS workloads
Whoishostingthis: Various SQL resources
StorageAcceleration: What, When, Why & How to Accelerate Storage
Filesystems.org: Various tools and links
StorageIOblog: Can we get a side of context with them IOPS and other storage metrics?

flash ssd and hdd

BrightTalk Webinar: Data Center Monitoring – Metrics that Matter for Effective Management
StorageIOblog: Enterprise SSHD and Flash SSD Part of an Enterprise Tiered Storage Strategy
StorageIOblog: Has SSD put Hard Disk Drives (HDD’s) On Endangered Species List?

server storage I/O bottlenecks and I/O blender

Microsoft TechNet: Measuring Disk Latency with Windows Performance Monitor (Perfmon)
Via Scalegrid.io: How to benchmark MongoDB with YCSB? (Perfmon)
Microsoft MSDN: List of Perfmon counters for sql server
Microsoft TechNet: Taking Your Server’s Pulse
StorageIOblog: Part II: How many IOPS can a HDD, HHDD or SSD do with VMware?
CMG: I/O Performance Issues and Impacts on Time-Sensitive Applications

flash ssd and hdd

Virtualization Practice: IO IO it is off to Storage and IO metrics we go
InfoStor: Is HP Short Stroking for Performance and Capacity Gains?
StorageIOblog: Is Computer Data Storage Complex? It Depends
StorageIOblog: More storage and IO metrics that matter
StorageIOblog: Moving Beyond the Benchmark Brouhaha
Yellow-Bricks: VSAN VDI Benchmarking and Beta refresh!

server storage I/O benchmark example

YellowBricks: VSAN performance: many SAS low capacity VS some SATA high capacity?
YellowBricsk: VSAN VDI Benchmarking and Beta refresh!
StorageIOblog: Seagate 1200 12Gbs Enterprise SAS SSD StorgeIO lab review
StorageIOblog: Part II: Seagate 1200 12Gbs Enterprise SAS SSD StorgeIO lab review
StorageIOblog: Server Storage I/O Network Benchmark Winter Olympic Games

flash ssd and hdd

VMware VDImark aka View Planner (also here, here and here) as well as VMmark here
StorageIOblog: SPC and Storage Benchmarking Games
StorageIOblog: Speaking of speeding up business with SSD storage
StorageIOblog: SSD and Storage System Performance

Hadoop server storage I/O performance
Various Server Storage I/O tools in a hadoop environment

Michael-noll.com: Benchmarking and Stress Testing an Hadoop Cluster With TeraSort, TestDFSIO
Virtualization Practice: SSD options for Virtual (and Physical) Environments Part I: Spinning up to speed on SSD
StorageIOblog: Storage and IO metrics that matter
InfoStor: Storage Metrics and Measurements That Matter: Getting Started
SilvertonConsulting: Storage throughput vs. IO response time and why it matters
Splunk: The percentage of Read / Write utilization to get to 800 IOPS?

flash ssd and hdd
Various server storage I/O benchmarking tools

Spiceworks: What is the best IO IOPs testing tool out there
StorageIOblog: How many IOPS can a HDD, HHDD or SSD do?
StorageIOblog: Some Windows Server Storage I/O related commands
Openmaniak: Iperf overview and Iperf.fr: Iperf overview
StorageIOblog: Server and Storage I/O Benchmark Tools: Microsoft Diskspd (Part I and Part II)
Quest: SQL Server Perfmon Poster (PDF)
Server and Storage I/O Networking Performance Management (webinar)
Data Center Monitoring – Metrics that Matter for Effective Management (webinar)
Flash back to reality – Flash SSD Myths and Realities (Industry trends & benchmarking tips), (MSP CMG presentation)
DBAstackexchange: How can I determine how many IOPs I need for my AWS RDS database?
ITToolbox: Benchmarking the Performance of SANs

server storage IO labs

StorageIOblog: Dell Inspiron 660 i660, Virtual Server Diamond in the rough (Server review)
StorageIOblog: Part II: Lenovo TS140 Server and Storage I/O Review (Server review)
StorageIOblog: DIY converged server software defined storage on a budget using Lenovo TS140
StorageIOblog: Server storage I/O Intel NUC nick knack notes First impressions (Server review)
StorageIOblog & ITKE: Storage performance needs availability, availability needs performance
StorageIOblog: Why SSD based arrays and storage appliances can be a good idea (Part I)
StorageIOblog: Revisiting RAID storage remains relevant and resources

Interested in cloud and object storage visit our objectstoragecenter.com page, for flash SSD checkout storageio.com/ssd page, along with data protection, RAID, various industry links and more here.

Additional learning experiences along with common questions (and answers), as well as tips can be found in Software Defined Data Infrastructure Essentials book.

Software Defined Data Infrastructure Essentials Book SDDC

What This All Means

Watch for additional links to be added above in addition to those that appear via comments.

Ok, nuff said, for now.

Gs

Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

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.

Posted on

Seagate 1200 12Gbs Enterprise SAS SSD StorgeIO lab review

Seagate 1200 12Gbs Enterprise SAS SSD StorgeIO lab review

This is the first post of a two part series, read the second post here.

Earlier this year I had the opportunity to test drive some Seagate 1200 12Gbs Enterprise SAS SSD’s as a follow-up to some earlier activity trying their Enterprise TurboBoost Drives. Disclosure: Seagate has been a StorageIO client and was also the sponsor of this white paper and associated proof-points mentioned in this post.

The question to ask yourself is not if flash Solid State Device (SSD) technologies are in your future, Instead the questions are when, where, using what, how to configure and related themes. SSD including traditional DRAM and NAND flash-based technologies are like real estate where location matters; however, there are different types of properties to meet various needs. This means leveraging different types of NAND flash SSD technologies in different locations in a complementary and cooperative aka hybrid way. For example nand flash SSD as part of an enterprise tiered storage strategy can be implemented server-side using PCIe cards, SAS and SATA drives as targets or as cache along with software, as well as leveraging SSD devices in storage systems or appliances.

Seagate 1200 SSD
Seagate 1200 Enterprise SAS 12Gbs SSD Image via Seagate.com

Another place where nand flash can be found and compliments SSD devices are so-called Solid State Hybrid Drives (SSHD) or Hybrid Hard Disk Drives (HHDD) including a new generation that accelerate writes as well as reads such as those Seagate refers to as with Enterprise TurboBoost. The Enterprise TurboBoost drives (view the companion StorageIO Lab review TurboBoost white paper here) were previously known as the Solid State Hybrid Drives (SSHD) or Hybrid Hard Disk Drives (HHDD). Read more about TurboBoost here and here.

The best server and storage I/O is the one you do not have to do

Keep in mind that the best server or storage I/O is that one that you do not have to do, with the second best being the one with the least overhead resolved as close to the processor (compute) as possible or practical. The following figure shows that the best place to resolve server and storage I/O is as close to the compute processor as possible however only a finite amount of storage memory located there. This is where the server memory and storage I/O hierarchy comes into play which is also often thought of in the context of tiered storage balancing performance and availability with cost and architectural limits.

Also shown is locality of reference which refers to how close data is to where it is being used and includes cache effectiveness or buffering. Hence a small amount of cache of flash and DRAM in the right location can have a large benefit. Now if you can afford it, install as much DRAM along with flash storage as possible, however if you are like most organizations with finite budgets yet server and storage I/O challenges, then deploy a tiered flash storage strategy.

flash cache locality of reference
Server memory storage I/O hierarchy, locality of reference

Seagate 1200 12Gbs Enterprise SAS SSD’s

Back to the Seagate 1200 12Gbs Enterprise SAS SSD which is covered in this StorageIO Industry Trends Perspective thought leadership white paper. The focus of the white paper is to look at how the Seagate 1200 Enterprise class SSD’s and 12Gbps SAS address current and next generation tiered storage for virtual, cloud, traditional Little and Big Data infrastructure environments.

Seagate 1200 Enteprise SSD

This includes providing proof points running various workloads including Database TPC-B, TPC-E and Microsoft Exchange in the StorageIO Labs along with cache software comparing SSD, SSHD and different HDD’s including 12Gbs SAS 6TB near-line high-capacity drives.

Seagate 1200 Enterprise SSD Proof Points

The proof points in this white paper are from an applications focus perspective representing more of an end-to-end real-world situation. While they are not included in this white paper, StorageIO has run traditional storage building-block focus workloads, which can be found at StorageIOblog (Part II: How many IOPS can a HDD, HHDD or SSD do with VMware?). These include tools such as Iometer, iorate, vdbench among others for various IO sizes, mixed, random, sequential, reads, writes along with “hot-band" across different number of threads (concurrent users). “Hot-Band” is part of the SNIA Emerald energy effectiveness metrics for looking at sustained storage performance using tools such as vdbench. Read more about other various server and storage I/O benchmarking tools and techniques here.

For the following series of proof-points (TPC-B, TPC-E and Exchange) a system under test (SUT) consisted of a physical server (described with the proof-points) configured with VMware ESXi along with guests virtual machines (VMs) configured to do the storage I/O workload. Other servers were used in the case of TPC workloads as application transactional requester to drive the SQL Server database and resulting server storage I/O workload. VMware was used in the proof-points to reflect a common industry trend of using virtual server infrastructures (VSI) supporting applications including database, email among others. For the proof-point scenarios, the SUT along with storage system device under test were dedicated to that scenario (e.g. no other workload running) unless otherwise noted.

Server Storage I/O config
Server Storage I/O configuration for proof-points

Microsoft Exchange Email proof-point configuration

For this proof-point, Microsoft Jet Stress Exchange performance workloads were placed (e.g. Exchange Database – EDB file) on each of the different devices under test with various metrics shown including activity rates and response time for reads as well as writes. For the Exchange testing, the EDB was placed on the device being tested while its log files were placed on a separate Seagate 400GB Enterprise 12Gbps SAS SSD.

Test configuration: Seagate 400GB 12000 2.5” SSD (ST400FM0073) 12Gbps SAS, 600GB 2.5” Enterprise 15K with TurboBoost™ (ST600MX) 6 Gbps SAS, 600GB 2.5” Enterprise Enhanced 15K V4 (15K RPM) HDD (ST600MP) with 6 Gbps SAS, Seagate Enterprise Capacity Nearline (ST6000NM0014) 6TB 3.5” 7.2K RPM HDD 12 Gbps SAS and 3TB 7.2K SATA HDD. Email server hosted as guest on VMware vSphere/ESXi V5.5, Microsoft SBS2011 Service Pack 1 64 bit. Guest VM (VMware vSphere 5.5) was on a SSD based dat, had a physical machine (host), with 14 GB DRAM, quad CPU (4 x 3.192GHz) Intel E3-1225 v300, with LSI 9300 series 12Gbps SAS adapters in a PCIe Gen 3 slot with Jet Stress 2010.  All devices being tested were Raw Device Mapped (RDM) where EDB resided. VM on a SSD based separate data store than devices being tested. Log file IOPs were handled via a separate SSD device also persistent (no delayed writes). EDB was 300GB and workload ran for 8 hours.

Microsoft Exchange VMware SSD performance
Microsoft Exchange proof-points comparing various storage devices

TPC-B (Database, Data Warehouse, Batch updates) proof-point configuration

SSD’s are a good fit for both transaction database activity with reads and write as well as query-based decision support systems (DSS), data warehouse and big data analytics. The following are proof points of SSD capabilities for database activity. In addition to supporting database table files and objects, along with transaction journal logs, other uses include for meta-data, import/export or other high-IO and write intensive scenarios. Two database workload profiles were tested including batch update (write-intensive) and transactional. Activity involved running Transaction Performance Council (TPC) workloads TPC-B (batch update) and TPC-E (transaction/OLTP simulate financial trading system) against Microsoft SQL Server 2012 databases. Each test simulation had the SQL Server database (MDF) on a different device with transaction log file (LDF) on a separate SSD. TPC-B for a single device results shown below.

TPC-B (write intensive) results below show how TPS work being done (blue) increases from left to right (more is better) for various numbers of simulated users. Also shown on the same line for each amount of TPS work being done is the average latency in seconds (right to left) where lower is better. Results are shown from top to bottom for each group of users (100, 50, 20 and 1) for the different drives being tested (top to bottom). Note how the SSD device does more work at a lower response time vs. traditional HDD’s

Test configuration: Seagate 400GB 12000 2.5” SSD (ST400FM0073) 12Gbps SAS, 600GB 2.5” Enterprise 15K with TurboBoost™ (ST600MX) 6 Gbps SAS, 600GB 2.5” Enterprise Enhanced 15K V4 (15K RPM) HDD (ST600MP) with 6 Gbps SAS, Seagate Enterprise Capacity Nearline (ST6000NM0014) 6TB 3.5” 7.2K RPM HDD 12 Gbps SAS and 3TB Seagate 7.2K SATA HDD Workload generator and virtual clients Windows 7 Ultimate 64 bit. Microsoft SQL Server 2012 database was on Windows 7 guest. Guest VM (VMware vSphere 5.5) had a dedicated 14 GB DRAM, quad CPU (4 x 3.192GHz) Intel E3-1225 v300, with LSI 9300 series 12Gbps SAS adapters in a PCIe Gen 3 slot along with TPC-B (www.tpc.org) workloads.

VM with guest OS along with SQL tempdb and masterdb resided on separate SSD based data store from devices being tested (e.g., where MDF (main database tables) and LDF (log file) resided). All devices being tested were Raw Device Mapped (RDM) independent persistent with database log file on a separate SSD device also persistent (no delayed writes) using VMware PVSCSI driver. MDF and LDF file sizes were 142GB and 26GB with scale factor of 10000, with each step running for one hour (10-minute preamble). Note that these proof-points DO NOT use VMware or any other third-party cache software or I/O acceleration tool technologies as those are covered later in a separate proof-point.

TPC-B sql server database SSD performance
TPC-B SQL Server database proof-points comparing various storage devices

TPC-E (Database, Financial Trading) proof-point configuration

The following shows results from TPC-E test (OLTP/transactional workload) simulating a financial trading system. TPC-E is an industry standard workload that performs a mix of reads and writes database queries. Proof-points were performed with various numbers of users from 10, 20, 50 and 100 to determine (TPS) Transaction per Second (aka I/O rate) and response time in seconds. The TPC-E transactional results are shown for each device being tested across different user workloads. The results show how TPC-E TPS work (blue) increases from left to right (more is better) for larger numbers of users along with corresponding latency (green) that goes from right to left (less is better). The Seagate Enterprise 1200 SSD is shown on the top in the figure below with a red box around its results. Note how the SSD as a lower latency while doing more work compared to the other traditional HDD’s

Test configuration: Seagate 400GB 12000 2.5” SSD (ST400FM0073) 12Gbps SAS, 600GB 2.5” Enterprise 15K with TurboBoost™ (ST600MX) 6 Gbps SAS, 600GB 2.5” Enterprise Enhanced 15K V4 (15K RPM) HDD (ST600MP) with 6 Gbps SAS, Seagate Enterprise Capacity Nearline (ST6000NM0014) 6TB 3.5” 7.2K RPM HDD 12 Gbps SAS and 3TB Seagate 7.2K SATA HDD Workload generator and virtual clients Windows 7 Ultimate 64 bit. Microsoft SQL Server 2012 database was on Windows 7 guest. Guest VM (VMware vSphere 5.5) had a dedicated 14 GB DRAM, quad CPU (4 x 3.192GHz) Intel E3-1225 v300, with LSI 9300 series 12Gbps SAS adapters in a PCIe Gen 3 slot along with TPC-B (www.tpc.org) workloads.

VM with guest OS along with SQL tempdb and masterdb resided on separate SSD based data store from devices being tested (e.g., where MDF (main database tables) and LDF (log file) resided). All devices being tested were Raw Device Mapped (RDM) independent persistent with database log file on a separate SSD device also persistent (no delayed writes) using VMware PVSCSI driver. MDF and LDF file sizes were 142GB and 26GB with scale factor of 10000, with each step running for one hour (10-minute preamble). Note that these proof-points DO NOT use VMware or any other third-party cache software or I/O acceleration tool technologies as those are covered later in a separate proof-point.

TPC-E sql server database SSD performance
TPC-E (Financial trading) SQL Server database proof-points comparing various storage devices

Continue reading part-two of this two-part series here including the virtual server storage I/O blender effect and solution.

Ok, nuff said (for now).

Cheers gs

Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press) and Resilient Storage Networks (Elsevier)
twitter @storageio

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved

Posted on

Enterprise SSHD and Flash SSD Part of an Enterprise Tiered Storage Strategy

Enterprise SSHD and Flash SSD Part of an Enterprise Tiered Storage Strategy

The question to ask yourself is not if flash Solid State Device (SSD) technologies are in your future.

Instead the questions are when, where, using what, how to configure and related themes. SSD including traditional DRAM and NAND flash-based technologies are like real estate where location matters; however, there are different types of properties to meet various needs. This means leveraging different types of NAND flash SSD technologies in different locations in a complementary and cooperative aka hybrid way.

Introducing Solid State Hybrid Drives (SSHD)

Solid State Hybrid Disks (SSHD) are the successors to previous generation Hybrid Hard Disk Drives (HHDD) that I have used for several years (you can read more about them here, and here).

While it would be nice to simply have SSD for everything, there are also economic budget realities to be dealt with. Keep in mind that a bit of nand flash SSD cache in the right location for a given purpose can go a long way which is the case with SSHDs. This is also why in many environments today there is a mix of SSD, HDD of various makes, types, speeds and capacities (e.g. different tiers) to support diverse application needs (e.g. not everything in the data center is the same).

However, If you have the need for speed and can afford or benefit from the increased productivity by all means go SSD!

Otoh if you have budget constraints and need more space capacity yet want some performance boost, then SSHDs are an option. The big difference however between today’s SSHDs that are available for both enterprise class storage systems and servers, as well as desktop environments is that they can accelerate both reads and writes. This is different from their predecessors that I have used for several years now that had basic read acceleration, however no write optimizations.

SSHD storage I/O oppourtunity
Better Together: Where SSHDs fit in an enterprise tiered storage environment with SSD and HDDs

As their names imply, they are a hybrid between a nand flash Solid State Device (SSD) and traditional Hard Disk Drive (HDD) meaning a best of situation. This means that the SSHD are based on a traditional spinning HDD (various models with different speeds, space capacity, interfaces) along with DRAM (which is found on most modern HDDs), along with nand flash for read cache, and some extra nonvolatile memory for persistent write cache combined with a bit of software defined storage performance optimization algorithms.

Btw, if you were paying attention to that last sentence you would have picked up on something about nonvolatile memory being used for persistent write cache which should prompt the question would that help with nand flash write endurance? Yup.

Where and when to use SSHD?

In the StorageIO Industry Trends Perspective thought leadership white paper I recently released compliments of Seagate Enterprise Turbo SSHD (that’s a disclosure btw ;) enterprise class Solid State Hybrid Drives (SSHD) were looked at and test driven in the StorageIO Labs with various application workloads. These activities include being in a virtual environment for common applications including database and email messaging using industry standard benchmark workloads (e.g. TPC-B and TPC-E for database, JetStress for Exchange).

Storage I/O sshd white paper

Conventional storage system focused workloads using iometer, iorate and vdbench were also run in the StorageIO Labs to set up baseline reads, writes, random, sequential, small and large I/O size with IOPs, bandwidth and response time latency results. Some of those results can be found here (Part II: How many IOPS can a HDD, HHDD or SSD do with VMware?) with other ongoing workloads continuing in different configurations. The various test drive proof points were done in the   comparing SSHD, SSD and different HDDs.

Data Protection (Archiving, Backup, BC, DR)

Staging cache buffer area for snapshots, replication or current copies before streaming to other storage tier using fast read/write capabilities. Meta data, index and catalogs benefit from fast reads and writes for faster protection.

Big Data DSS
Data Warehouse

Support sequential read-ahead operations and “hot-band” data caching in a cost-effective way using SSHD vs. slower similar capacity size HDDs for Data warehouse, DSS and other analytic environments.

Email, Text and Voice Messaging

Microsoft Exchange and other email journals, mailbox or object repositories can leverage faster read and write I/Os with more space capacity.

OLTP, Database
 Key Value Stores SQL and NoSQL

Eliminate the need to short stroke HDDs to gain performance, offer more space capacity and IOP performance per device for tables, logs, journals, import/export and scratch, temporary ephemeral storage. Leverage random and sequential read acceleration to compliment server-side SSD-based read and write-thru caching. Utilize fast magnetic media for persistent data reducing wear and tear on more costly flash SSD storage devices.

Server Virtualization

Fast disk storage for data stores and virtual disks supporting VMware vSphere/ESXi, Microsoft Hyper-V, KVM, Xen and others.  Holding virtual machines such as VMware VMDKs, along with Hyper-V and other hypervisor virtual disks.  Compliment virtual server read cache and I/O optimization using SSD as a cache with writes going to fast SSHD. For example VMware V5.5 Virtual SAN host disk groups use SSD as a read cache and can use SSHD as the magnetic disk for storing data while boosting performance without breaking the budget or adding complexity.

Speaking of Virtual, as mentioned the various proof points were run using Windows systems that were VMware guests with the SSHD and other devices being Raw Device Mapped (RDM) SAS and SATA attached, read how to do that here.

Hint: If you know about the VMware trick for making a HDD look like a SSD to vSphere/ESXi (refer to here and here) think outside the virtual box for a moment on some things you could do with SSHD in a VSAN environment among other things, for now, just sayin ;).

Virtual Desktop Infrastructure (VDI)

SSHD can be used as high performance magnetic disk for storing linked clone images, applications and data. Leverage fast read to support read ahead or pre-fetch to compliment SSD based read cache solutions. Utilize fast writes to quickly store data enabling SSD-based read or write-thru cache solutions to be more effective. Reduce impact of boot, shutdown, and virus scan or maintenance storms while providing more space capacity.

Table 1 Example application and workload scenarios benefiting from SSHDs

Test drive application proof points

Various workloads were run using Seagate Enterprise Turbo SSHD in the StorageIO lab environment across different real world like application workload scenarios. These include general storage I/O performance characteristics profiling (e.g. reads, writes, random, sequential or various IOP size) to understand how these devices compare to other HDD, HHDD and SSD storage devices in terms of IOPS, bandwidth and response time (latency). In addition to basic storage I/O profiling, the Enterprise Turbo SSHD was also used with various SQL database workloads including Transaction Processing Council (TPC); along with VMware server virtualization among others use case scenarios.

Note that in the following workload proof points a single drive was used meaning that using more drives in a server or storage system should yield better performance. This also means scaling would be bound by the constraints of a given configuration, server or storage system. These were also conducted using 6Gbps SAS with PCIe Gen 2 based servers and ongoing testing is confirming even better results with 12Gbs SAS, faster servers with PCIe Gen 3.

SSHD large file storage i/o
Copy (read and write) 80GB and 220GB file copies (time to copy entire file)

SSHD storage I/O TPCB Database performance
SQLserver TPC-B batch database updates

Test configuration: 600GB 2.5” Enterprise Turbo SSHD (ST600MX) 6 Gbps SAS, 600GB 2.5” Enterprise Enhanced 15K V4 (15K RPM) HDD (ST600MP) with 6 Gbps SAS, 500GB 3.5” 7.2K RPM HDD 3 Gbps SATA, 1TB 3.5” 7.2K RPM HDD 3 Gbps SATA. Workload generator and virtual clients ran on Windows 7 Ultimate. Microsoft SQL Server 2012 Database was on Windows 7 Ultimate SP1 (64 bit) 14 GB DRAM, Dual CPU (Intel x3490 2.93 GHz)), with LSI 9211 6Gbps SAS adapters with TPC-B (www.tpc.org) workloads. VM resided on separate data store from devices being tested. All devices being tested with SQL MDF were Raw Device Mapped (RDM) independent persistent with database log file (LDF) on a separate SSD device also persistent (no delayed writes). Tests were performed in StorageIO Lab facilities by StorageIO personal.

SSHD storage I/O TPCE Database performance
SQLserver TPC-E transactional workload

Test configuration: 600GB 2.5” Enterprise Turbo SSHD (ST600MX) 6 Gbps SAS, 600GB 2.5” Enterprise Enhanced 15K V4 (15K RPM) HDD (ST600MP) with 6 Gbps SAS, 300GB 2.5” Savio 10K RPM HDD 6 Gbps SAS, 1TB 3.5” 7.2K RPM HDD 6 Gbps SATA. Workload generator and virtual clients Windows 7 Ultimate. Microsoft SQL Server 2012 database was on Windows 7 Ultimate SP1 (64 bit) 14 GB DRAM, Dual CPU (E8400 2.99GHz), with LSI 9211 6Gbps SAS adapters with TPC-E (www.tpc.org) workloads. VM resided on separate SSD based data store from devices being tested (e.g., where MDF resided). All devices being tested were Raw Device Mapped (RDM) independent persistent with database log file on a separate SSD device also persistent (no delayed writes). Tests were performed in StorageIO Lab facilities by StorageIO personal.

SSHD storage I/O Exchange performance
Microsoft Exchange workload

Test configuration: 2.5” Seagate 600 Pro 120GB (ST120FP0021 ) SSD 6 Gbps SATA, 600GB 2.5” Enterprise Turbo SSHD (ST600MX) 6 Gbps SAS, 600GB 2.5” Enterprise Enhanced 15K V4 (15K RPM) HDD (ST600MP) with 6 Gbps SAS, 2.5” Savio 146GB HDD 6 Gbps SAS, 3.5” Barracuda 500GB 7.2K RPM HDD 3 Gbps SATA. Email server hosted as guest on VMware vSphere/ESXi V5.5, Microsoft Small Business Server (SBS) 2011 Service Pack 1 64 bit, 8GB DRAM, One CPU (Intel X3490 2.93 GHz) LSI 9211 6 Gbps SAS adapter, JetStress 2010 (no other active workload during test intervals). All devices being tested were Raw Device Mapped (RDM) where EDB resided. VM on a SSD based separate data store than devices being tested. Log file IOPs were handled via a separate SSD device.

Read more about the above proof points along view data points and configuration information in the associated white paper found here (no registration required).

What this all means

Similar to flash-based SSD technologies the question is not if, rather when, where, why and how to deploy hybrid solutions such as SSHDs. If your applications and data infrastructures environment have the need for storage I/O speed without loss of space capacity and breaking your budget, SSD enabled devices like the Seagate Enterprise Turbo 600GB SSHD are in your future. You can learn more about enterprise class SSHD such as those from Seagate by visiting this link here.

Watch for extra workload proof points being performed including with 12Gbps SAS and faster servers using PCIe Gen 3.

Ok, nuff said.

Cheers
Gs

Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press) and Resilient Storage Networks (Elsevier)
twitter @storageio

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved

Posted on

EPA Energy Star for Data Center Storage Update

EPA Energy Star

Following up on a recent post about Green IT, energy efficiency and optimization for servers, storage and more, here are some additional  thoughts, perspectives along with industry activity around the U.S. Environmental Protection Agency (EPA) Energy Star for Server, Data Center Storage and Data Centers.

First a quick update, Energy Star for Servers is in place with work now underway on expanding and extending beyond the first specification. Second is that Energy Star for Data Center storage definition is well underway including a recent workshop to refine the initial specification along with discussion for follow-on drafts.

Energy Star for Data Centers is also currently undergoing definition which is focused more on macro or facility energy (notice I did not say electricity) efficiency as opposed to productivity or effectiveness, items that the Server and Storage specifications are working towards.

Among all of the different industry trade or special interests groups, at least on the storage front the Storage Networking Industry Association (SNIA) Green Storage Initiative (GSI) and their Technical Work Groups (TWG) have been busily working for the past couple of years on taxonomies, metrics and other items in support of EPA Energy Star for Data Center Storage.

A challenge for SNIA along with others working on related material pertaining to storage and efficiency is the multi-role functionality of storage. That is, some storage simply stores data with little to no performance requirements while other storage is actively used for reading and writing. In addition, there are various categories, architectures not to mention hardware and software feature functionality or vendors with different product focus and interests.

Unlike servers that are either on and doing work, or, off or in low power mode, storage is either doing active work (e.g. moving data), storing in-active or idle data, or a combination of both. Hence for some, energy efficiency is about how much data can be stored in a given footprint with the least amount of power known as in-active or idle measurement.

On the other hand, storage efficiency is also about using the least amount of energy to produce the most amount of work or activity, for example IOPS or bandwidth per watt per footprint.

Thus the challenge and need for at least a two dimensional  model looking at, and reflecting different types or categories of storage aligned for active or in-active (e.g. storing) data enabling apples to apples, vs. apples to oranges comparison.

This is not all that different from how EPA looks at motor vehicle categories of economy cars, sport utility, work or heavy utility among others when doing different types of work, or, in idle.

What does this have to do with servers and storage?

Simple, when a server powers down where does its data go? That’s right, to a storage system using disk, ssd (RAM or flash), tape or optical for persistency. Likewise, when there is work to be done, where does the data get read into computer memory from, or written to? That’s right, a storage system. Hence the need to look at storage in a multi-tenant manner.

The storage industry is diverse with some vendors or products focused on performance or activity, while others on long term, low cost persistent storage for archive, backup, not to mention some doing a bit of both. Hence the nomenclature of herding cats towards a common goal when different parties have various interests that may conflict yet support needs of various customer storage usage requirements.

Figure 1 shows a simplified, streamlined storage taxonomy that has been put together by SNIA representing various types, categories and functions of data center storage. The green shaded areas are a good step in the right direction to simplify yet move towards realistic and achievable befits for storage consumers.


Figure 1 Source: EPA Energy Star for Data Center Storage web site document

The importance of the streamlined SNIA taxonomy is to help differentiate or characterize various types and tiers of storage (Figure 2) products facilitating apples to apples comparison instead of apples or oranges. For example, on-line primary storage needs to be looked at in terms of how much work or activity per energy footprint determines efficiency.


Figure 2: Tiered Storage Example

On other hand, storage for retaining large amounts of data that is in-active or idle for long periods of time should be looked at on a capacity per energy footprint basis. While final metrics are still being flushed out, some examples could be active storage gauged by IOPS or work or bandwidth per watt of energy per footprint while other storage for idle or inactive data could be looked at on a capacity per energy footprint basis.

What benchmarks or workloads to be used for simulating or measuring work or activity are still being discussed with proposals coming from various sources. For example SNIA GSI TWG are developing measurements and discussing metrics, as have the storage performance council (SPC) and SPEC among others including use of simulation tools such as IOmeter, VMware VMmark, TPC, Bonnie, or perhaps even Microsoft ESRP.

Tenants of Energy Star for Data Center Storage overtime hopefully will include:

  • Reflective of different types, categories, price-bands and storage usage scenarios
  • Measure storage efficiency for active work along with in-active or idle usage
  • Provide insight for both storage performance efficiency and effective capacity
  • Baseline or raw storage capacity along with effective enhanced optimized capacity
  • Easy to use metrics with more in-depth back ground or disclosure information

Ultimately the specification should help IT storage buyers and decision makers to compare and contrast different storage systems that are best suited and applicable to their usage scenarios.

This means measuring work or activity per energy footprint at a given capacity and data protection level to meet service requirements along with during in-active or idle periods. This also means showing storage that is capacity focused in terms of how much data can be stored in a given energy footprint.

One thing that will be tricky however will be differentiating GBytes per watt in terms of capacity, or, in terms of performance and bandwidth.

Here are some links to learn more:

Stay tuned for more on Energy Star for Data Centers, Servers and Data Center Storage.

Ok, nuff said.

Cheers gs

Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press) and Resilient Storage Networks (Elsevier)
twitter @storageio

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved

Posted on

SPC and Storage Benchmarking Games

Storage I/O trends

There is a post over in one of the LinkedIn Discussion forums about storage performance council (SPC) benchmarks being miss-leading that I just did a short response post to. Here’s the full post as LinkedIn has a short post response limit.

While the SPC is far from perfect, it is at least for block, arguably better than doing nothing.

For the most part, SPC has become a de facto standard for at least block storage benchmarks independent of using IOmeter or other tools or vendor specific simulations, similar how MSFT ESRP is for exchange, TPC for database, SPEC for NFS and so forth. In fact, SPC even recently rather quietly rolled out a new set of what could be considered the basis for Green storage benchmarks. I would argue that SPC results in themselves are not misleading, particularly if you take the time to look at both the executive and full disclosures and look beyond the summary.

Some vendors have taken advantage of the SPC results playing games with discounting on prices (something that’s allowed under SPC rules) to show and make apples to oranges comparisons on cost per IOP or other ploys. This proactive is nothing new to the IT industry or other industries for that matter, hence benchmark games.

Where the misleading SPC issue can come into play is for those who simply look at what a vendor is claiming and not looking at the rest of the story, or taking the time to look at the results and making apples to apples, instead of believing the apples to oranges comparison. After all, the results are there for a reason. That reason is for those really interested to dig in and sift through the material, granted not everyone wants to do that.

For example, some vendors can show a highly discounted list price to get a better IOP per cost on an apple to oranges basis, however, when processes are normalized, the results can be quite different. However here’s the real gem for those who dig into the SPC results, including looking at the configurations and that is that latency under workload is also reported.

The reason that latency is a gem is that generally speaking, latency does not lie.

What this means is that if vendor A doubles the amount of cache, doubles the number of controllers, doubles the number of disk drives, plays games with actual storage utilization (ASU), utilizes fast interfaces from 10 GbE  iSCSI to 8Gb FC or FCoE or SAS to get a better cost per IOP number with discounting, look at the latency numbers. There have been some recent examples of this where vendor A has a better cost per IOP while achieving a higher number of IOPS at a lower cost compared to vendor B, which is what is typically reported in a press release or news story. (See a blog entry that also points to a CMG presentation discussion around this topic here.

Then go and look at the two results, vendor B may be at list price while vendor A is severely discounted which is not a bad thing, as that is then the starting list price as to which customers should start negotiations. However to be fair, normalize the pricing for fun, look at how much more equipment vendor A may need while having to discount to get the price to offset the increased amount of hardware, then look at latency.

In some of the recent record reported results, the latency results are actually better for a vendor B than for a vendor A and why does latency matter? Beyond showing what a controller can actually do in terms of levering  the number of disks, cache, interface ports and so forth, the big kicker is for those talking about SSD (RAM or FLASH) in that SSD generally is about latency. To fully effectively utilize SSD which is a low latency device, you would want a controller that can do a decent job at handling IOPS; however you also need a controller that can do a decent job of handling IOPS with low latency under heavy workload conditions.

Thus the SPC again while far from perfect, at least for a thumb nail sketch and comparison is not necessarily misleading, more often than not it’s how the results are utilized that is misleading. Now in the quest for the SPC administrators to try and gain more members and broader industry participation and thus secure their own future, is the SPC organization or administration opening itself up to being used more and more as a marketing tool in ways that potentially compromise all the credibility (I know, some will dispute the validity of SPC, however that’s reserved for a different discussion ;) )?

There is a bit of Déjà here for those involved with RAID and storage who recall how the RAID Advisory Board (RAB) in its quest to gain broader industry adoption and support succumbed to marketing pressures and use or what some would describe as miss-use and is now a member of the “Where are they now” club!

Don’t get me wrong here; I like the SPC tests/results/format, there is a lot of good information in the SPC. The various vendor folks who work very hard behind the scenes to make the SPC actually work and continue to evolve it also all deserve a great big kudos, an “atta boy” or “atta girl” for the fine work that have been doing, work that I hope does not become lost in the quest to gain market adoption for the SPC.

Ok, so then this should all then beg the question of what is the best benchmark. Simple, the one that most closely resembles your actual applications, workload, conditions, configuration and environment.

Ok, nuff said.

Cheers gs

Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press) and Resilient Storage Networks (Elsevier)
twitter @storageio

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved