Share this post

July 28, 2015

SQL Server on Hyper-converged

Patrick Brennan - Atlantis

Avatar

The secret is out that enterprises are collecting more data than ever to provide the analytics organizations need to make critical business decisions.  In fact some reports indicate that data collection is now growing so rapidly that it’s doubling in size every two years.   These database driven activities provide vital information to the business for more efficient decision making, impacting their ability to compete, execute, and grow.  However, with the introduction of all of these new analytical capabilities, it presents a new set of challenges for an IT organization to deliver this enlightening information.  Not only does collecting all of this data consume a considerable amount of storage space, but it also takes a significant amount of time to analyze and process the data to deliver the answers an organization is looking for.

This brings up the greater point that the Atlantis USX hyper-converged solution enables organizations to finally take control of these challenges, freeing them from traditional storage systems that are not only slow and expensive, but also difficult to scale!  With Atlantis USX, businesses can make their datacenter to be more agile, allowing for scaling on demand with any hardware and storage the enterprise chooses to leverage.  Whether it is new or existing storage on DAS, NAS, SAN, Atlantis HyperScale, etc., Atlantis USX can provision an environment for immediate consumption while adding greater performance and reducing required capacity.

To further validate Atlantis’s ability to accelerate database performance, we conducted testing on the Atlantis USX platform with Microsoft SQL 2014.  In our findings, we were able to achieve a transaction per minute cost as low as $0.18 with hardware, software, and support included!  In addition, we were able to produce 5X more transactions per minute (TPM) than traditional storage systems.  As a result, organizations can reduce their batch job processing by more than 80%, giving access to critical information to organizations in a much shorter amount of time while spending less than half the cost of traditional servers with SAN!

SQL on Hyper-Converged Webinar: Cut your cost per transaction and SQL licensing in half!

To discover more about how Atlantis can make your datacenter more agile, respond and scale more quickly, and accelerate your database performance, please register for our webinar on August 13th at 9AM PT.  In this session, Jim Moyle and I will deep dive into the results of the performance testing he was able to perform on Atlantis USX and Microsoft SQL 2014 using HammerDB, SQLIO, and Iometer.  We'll also discuss how we're reducing SQL database capacity requirements by up to 80% with the deduplication technology built into Atlantis USX.

In the meantime, take a peek at the Atlantis USX Hyper-Converged Solution for Microsoft SQL 2014 Whitepaper.  This document covers more of the technical aspects on how Atlantis USX is alleviating the pain points of database workloads that continue to create headaches for enterprises.  I’ve included just a few excerpts of the high level findings we were able to achieve on just a single server!  You’ll find the server hardware configurations and other important details inside the whitepaper.

Atlantis USX Hyper-Converged Solution for Microsoft SQL 2014 Whitepaper

As Jim Moyle mentions in the whitepaper, the primary goals for most SQL administrators to consider when evaluating their database environment are: cost, performance, and resiliency.   Cost and performance share a very close relationship in SQL environments.  In order to lower costs, the performance of each server must be increased.  One method to measure this is by evaluating the amount of transactions per minute each SQL server is capable of delivering.  By increasing server efficiency, organizations can save considerable costs in database server licensing, storage, and hardware. Another method to measure performance is by evaluating the underlying storage subsystem as well.  To guarantee database performance isn’t limited by the underlying storage, the storage subsystem needs to be extremely fast and provide very low latency.  By introducing the Atlantis USX hyper-converged infrastructure, servers are able to produce more transactions per minute (TPM) and increase overall server efficiency throughout the datacenter.

Solution Price by Performance

As a result of the performance data collected during our testing, we were able to achieve significant cost savings when compared to traditional servers and SAN. 

Solution Pricing
Solution Pricing by Metric
TPM (Auto-Pilot) as low as: $0.18
SQLIO (4K Block)/IOPS as low as: $0.20
SQLIO (8K Block)/IOPS as low as: $0.21
SQLIO (64K Block)/IOPS as low as: $0.94
 

When leveraging Atlantis USX a TPM as low as $0.18 can be achieved as indicated above, even when you consider:

  • Server Hardware (with 3 year support)
  • Atlantis Software (with 3 year support)
  • Microsoft Windows Server 2012 R2 Standard
  • Microsoft SQL 2014 Standard Edition (8 vCPUs)
  • VMware vSphere 5.5 U2 (2 socket licensing)

Database Performance on Atlantis USX

With only two volumes provisioned, 76,650 IOPS were achieved using the Iometer testing tool with a SQL workload detailed in the whitepaper.  All tests conducted in the whitepaper were completed with only two volumes configured on a single host.  If greater performance is needed, additional volumes can be created to increase IOPS and throughput.  As a result, the incremental cost per IOPS is held to a minimum with the Atlantis USX platform. 

 
Iometer/HammerDB Test Results
Test Results
Iometer (IOPS) – 2 Volumes 76,650
HammerDB TPC-C Auto-Pilot (TPM) 319,036

 

When conducting the HammerDB TPC-C Auto-Pilot test, Atlantis USX continued to accelerate storage performance and achieved 319,036 TPM with 25 simultaneous workloads!  As each workload was added, the system continued to respond in a linear manner providing consistent and predictable scalability while maintaining a high level of performance.

 

SQLIO produced considerable numbers as well across the two volumes, again having additional capacity for more IOPS at your disposal with additional volumes.  In situations where 0ms was recorded, the latency was too low for the tool to measure due to performance Atlantis USX provides.  In all of the scenarios below, there would be no noticeable impact on performance despite the high number of IOPS and throughput being delivered by Atlantis USX.  This is just a small example of the performance that Atlantis USX can provide to your database infrastructures.
 

SQLIO Test Results
4K Block Size
Latency (in milliseconds, R/W) 0/0
IOPS (Read+Write) 122,137
Throughput (MB/s) 447
8K Block Size
Latency (in milliseconds, R/W) 0/0
IOPS (Read+Write) 115,084
Throughput (MB/s) 899
64K Block Size
Latency (in milliseconds, R/W) 0.5/0.5
IOPS (Read+Write) 33,437
Throughput (MB/s) 2089
 

Wrap Up

If organizations continue with traditional storage strategies, they will not only perpetuate an unsustainable cost model, but also fail to meet the demands of the business for better analytics and data.  Databases are growing exponentially in size with greater demands on performance than ever before.  By providing 5X more TPM at less than half the cost than traditional servers and SANs, Atlantis USX gives IT organizations the opportunity to spend less time fixing pain points, and more time innovating.

12345
Current rating: 5 (4 ratings)