What are you looking for ?
Infinidat
Articles_top

IDC Buyer Case Study: XtremIO

At $1.4 billion manufacturer of faucets

idc burgenerIDC Buyer Case Study: XtremIO (December 2015, IDC #US40631515) has been written by Eric Burgener, research director, storage, International Data Corp.

 

 

 

Manufacturer Drives Performance and Efficiency
by Deploying EMC XtremIO for its Mission-Critical SAP Environment

IDC opinion
As flash costs have come down, enterprises are leveraging the performance, reliability, and efficiency benefits it brings to the table to transform not only their IT infrastructure but also their businesses. All-flash arrays (AFAs) are one of the fastest-growing segments of the enterprise storage market, will generate a revenue of $2.53 billion in 2015, and should account for 60-70% of all primary storage spend by 2019. High data growth rates are challenging organizations to continue to deliver the performance necessary to meet internal and external customer expectations as businesses slowly move toward greater use of real-time analytics against extremely large data sets to plan and guide daily operations. A large, Midwestern supplier of kitchen and bath fixtures for commercial and residential use drives its business off of an SAP implementation that services a variety of both internal and external constituents on a daily basis. Performance is critical not only for interactive processes like sales and support but also for batch jobs that help feed production and distribution decisions. The company looked at flash technologies, integrated into existing enterprise storage arrays, very early on but, with an impending storage refresh in 2014 and concern about their ability to handle increasingly large data sets while meeting compliance requirements, it also looked at AFAs. EMC has been the company’s storage supplier of records for a number of years, and hence the company ended up purchasing two EMC XtremIO AFAs to host its mission-critical SAP environment.

Further:

  • The XtremIO AFA increased storage performance, enabling the company to complete batch jobs that fed planning processes up to nine times faster, thus allowing it to plan batch operations more efficiently and work with fresher and more complete data sets to drive better business decisions
  • AFA performance allowed the company to leverage a number of storage efficiency technologies like compression, thin provisioning, and space-efficient snapshots while continuing to meet or exceed internal (SLA targets, helping better manage data growth while at the same time lowering costs.
  • The XtremIO snapshot implementation has completely changed the company’s thinking on how to leverage copies to improve analytics, customer support, test and development, and data protection processes, improving both IT administrator and developer productivity to help lower costs and run operations more efficiently.

In this buyer case study
This IDC Buyer Case Study summarizes how a $1.4 billion manufacturer of faucets and other fixtures has integrated flash storage technology into its datacenter environment to improve performance, handle rapid business growth, and improve internal productivity. This study explores what drove the company’s initial interest in flash, how flash deployment has evolved in the company’s environment over time, what the company’s experience with flash has been, and what the company’s plans are for flash in the future.

Situation overview

Organization overview
The company, originally started in 1937, began as a faucet manufacturer and grew into a medium-sized business that manufactured and sold a line of faucets, sinks, showerheads, and related products for commercial and residential use. Several years ago, the company was acquired by a Fortune 1000 firm where it maintains its brand image as a subsidiary. The company is largely an EMC shop for enterprise storage, with installed VMAX and VNX systems. The company has a large SAP installation running manufacturing resource planning (MRP), customer relationship management (CRM), supply chain management (SCM), and a business warehouse, all of which are mission critical and run off Oracle databases.

As the company’s business grew over the years, the company always tracked storage performance to ensure that it continues to meet customer expectations, both internal and external to the company, with regard to application performance. The company’s manager of enterprise systems technologies first brought flash in as a tier in the company’s VMAX in 2010, using EMC’s automated storage tiering (FAST VP) to manage data placement across three tiers of storage media: flash, FC HDDs, and SATA HDDs. In preparation for a storage technology refresh in late 2014, he looked at the XtremIO all-flash array for the company’s SAP environment. Up until that point, the VMAX had been meeting the company’s performance requirements, but with expected data growth, he was concerned about being able to continue to meet performance expectations. The SAP environment ran entirely on dedicated physical servers, although the company had deployed some virtual infrastructure, based on VMware vSphere, running mixed file-system-based workloads against an EMC VNX2. The virtual workloads have historically been much less performance sensitive than their SAP environment.

Challenges and solution
In considering an AFA, the IT manager expected that he would get very high performance but was also interested in ease of use, inline compression, and a performant, scalable snapshot technology. He managed a small IT team with broad cross-functional responsibilities. Although he had senior administrative expertise on staff, all of his administrators were IT generalists, and he was trying to use automation as much as possible in datacenter workflows to maximize the productivity of his staff. The company had prior experience with deduplication technologies for use with secondary storage environments through its Data Domain array and knew that the feature brought little benefit with databases. So the manager was anxious to see what type of data reduction ratio an AFA’s inline compression might provide. The company’s existing workflows also depended heavily on high performance copies of databases, and there were issues with the cloning technology the company was currently using. Full clones took a long time to create, and if an administrator tried to mount a database in production. This was their first AFA, and it was a very easy sell to upper management based on the results they had seen.

Results
Based on the POC results, the company ended up purchasing and deploying two XtremIO arrays in early 2015, using EMC VPLEX Metro to replicate between the systems that were housed in different datacenters for DR purposes. VPLEX supports an active/active metro cluster configuration, supporting instant failover to the secondary site without any data loss or disruption. Each system included a single 20TB X-Brick, but each X-Brick is fully redundant, supporting automatic recovery from controller, SSD, power supply and fan failures, online replacement of all redundant components, and online firmware upgrades, meeting the stringent availability requirements in the company’s mission-critical SAP environment.

The POC performance and data reduction results were proven out in practice. The company employees who use the SAP system every day see a noticeable difference not only in response times but in particular batch job performance. Depending upon system load, it was sometimes difficult to get reports to complete, and they had to be submitted sometimes multiple times before they would run to completion. With XtremIO, batch jobs complete up to nine times faster than they did before, and they run to completion on the first submission. The three-weekend job mentioned previously can now be completed in one weekend, and faster completion for not only batch loads but analytics work allows employees to consistently work with fresher data, which allows them to make better, more efficient business decisions concerning production and distribution as well as customer-facing interaction. The ability to complete batch workloads so much quicker gives the IT manager and his team the flexibility to rearrange batch jobs when necessary to better meet business requirements.

The XtremIO GUI provides a way to track the average data reduction ratio the system is delivering, and since the system reached steady state (which was shortly after it was put into production in the spring of 2015) it has been delivering data reduction ratios in the 2.1-2.2 range. But XtremIO has brought other storage efficiency technologies to bear as well, including thin provisioning and space efficient snapshots and clones. XtremIO’s snapshot implementation, completely re-architected to take advantage of the scale-out architecture, memory-resident metadata, and flash media, has outperformed the company’s expectations. Read-only or R/W snapshots can be created instantly with no data movement and no storage capacity consumption. For break/fix purposes, developers are now able to respond more quickly, make more copies if/when they want to since space consumption is no longer an issue, and still meet SOX requirements. Using XtremIO’s REST API, the manager has been able to automate the process of snapshot creation against live data and the mounting of those snapshots onto the right hosts, ensuring that snapshot creation and usage is done reliably. This has saved his developers and technical support personnel many hours on a weekly basis, making them more productive and allowing them to work with very recent data. The ability to take snapshots of snapshots has allowed him to open and verify backups without losing them as a recovery point, and he is keeping longer snapshot histories online, reducing the odds that he would ever have to resort to Data Domain-based backups to have to recover from anything but a catastrophic event.

The IT manager has been extremely pleased with these first AFA purchases and plans to move as many primary applications to flash as he can going forward. The speed of migration will be determined by several factors, one of which is the technology refresh cycle for existing equipment (which is primarily leased). The company is currently doing a pilot on SAP HANA and will likely deploy that in production on XtremIO – even though HANA runs in memory, running it on a high-performance flash system ensures that customers can work with larger data sets that don’t fit in memory with almost no performance impacts (something that other customers have noted as a great reason to run HANA on an AFA as well). The company is also exploring virtualizing SAP whose four key systems today run on physical servers and is planning to stay on XtremIO for those systems even as it makes that move. The IT manager sees his company moving over time to an all-flash datacenter for primary storage but has no explicit strategy for it today – the company is letting business requirements drive flash deployments in the near term.

Essential guidance
The company has hosted four of its most mission-critical systems on an AFA, showing that it trusts its business to the XtremIO product. The move to an AFA was prompted by the company’s need to not only meet current performance requirements but also to be able to continue to meet those requirements – particularly in terms of latency – as the company’s business continues to grow. AFAs in general excel at delivering consistent performance, even under heavy loads – an important consideration as companies depend more on real-time interaction, real-time analytics, and real-time response to service both internal and external customer needs. Storage efficiency technologies like inline compression, deduplication, thin provisioning, and space-efficient snapshots help mitigate the cost impacts of high data growth environments even as they lower the effective dollar-per-gigabyte cost of AFAs, better enabling their use across a broader mix of application environments. What is most interesting about the company’s experience with AFAs to date is how a scalable, high performance snapshot implementation is allowing the company to evolve legacy workflows for improved response times, productivity, recovery reliability, and better business decisions. This highlights a point that IDC has made in the past, in particular about the XtremIO snapshot implementation: faced with its unique design, which leverages scale-out architecture, memory-resident metadata, and is built specifically for flash media, administrators must think out of the box about how they can best use snapshots to improve existing operations and workflows. By his own admission, the IT manager’s use of snapshots (which includes both read only and R/W) has improved exponentially since he implemented XtremIO, with very favorable results. When clones can be created instantly with no space consumption and retained and used with no performance impacts, people come up with all sorts of ways to use them to make their lives better. Ironically, he noted that an interesting side effect of being able to make so many high-performance SAP clones is that he now has to think about where to get the server resources to run the clones.

The company’s flash deployment experiences follow the approach of much of the industry to date. Initially experimenting with flash as a cache and then as a persistent tier in legacy arrays, performance requirements in high data growth environments led the company to consider an AFA. The company was very impressed with the performance difference, which was clearly noted by end users. The company has gotten some unexpected value from flash deployment as well in terms of business benefits, particularly by being able to leverage fresher data to improve business decisions and test/dev operations. And the company sees an ‘all flash for primary storage’ strategy as a future inevitability.

Articles_bottom
AIC
ATTO
OPEN-E