Hybrid arrays provide CIOs with a balance between performance and capacity

There are two hard facts that every CIO will face as some point as data growth continues:

  • Storage performance will become an exceedingly important metric, if it hasn’t already.
  • Flash storage alone will not solve an organizations storage problems due to the high $/TB cost.

As I state these two facts, I’m not including in the group of CIOs those that have what many consider to be niche needs.  Instead, I’m including those mainstream SMB and midmarket CIOs that run “real world” data centers.

Throw more spindles at the problem

It used to be the case that to solve a storage-related performance issue, a CIO had to throw more hardware at the problem in order to achieve the desired performance level.  Poor storage performance also led to such activities as [http://recoverymonkey.org/2012/07/24/what-is-hard-disk-short-stroking short stroking] as a way to gain additional performance at the cost of capacity.  In some way, though, CIOs were buying more hardware to solve what has become a more serious performance related trend.

Flash enters the scene, but…

When flash-based drives entered the scene, it appeared as if performance related issues would be solved once and for all.  However, whereas Flash-based solid state drives carried with them very low dollar per IOPS figures, the dollar per TB metric went the other way and into the clouds.  As such, CIOs could get great performance, but the cost was far beyond what most could afford to store in terms of overall capacity.

In short, the technology to solve the performance issue has been extremely expensive and has required a major trade off with regard to capacity.  In some instances, this might be ok, but in others, it’s not an acceptable compromise.

All flash arrays

In recent years, a number of startups have come on the scene with solutions that target the performance issues that have plagued some areas of IT.  While these solutions provide massive performance gains, their per-GB price is very high compared to rotational storage.  Many all-flash vendors have built comprehensive data reduction capabilities into their products in order to enable a lower per-GB cost, but they are still significantly more expensive than their rotational counterparts, although the performance is not even comparable since the all-flash solutions simply eviscerate the spinning rust solutions.

Hybrid arrays are the sweet spot

Many CIOs in the mainstream are looking for a better balance between performance and capacity than is possible with either pure rotational storage or pure flash storage.  That’s where hybrid storage comes in.

Hybrid storage arrays, such as those from Nimble, Tegile and Tintri, provide CIOs with a single storage device that provides significant capacity thanks to spinning storage while at the same time providing performance benefits thanks to the include solid state disks.  In working this way, the companies can keep their costs much lower than is possible with all-flash.

These vendors are also working hard on deduplication technologies in order to bring the realized costs of the solution down to spinning disk levels.  The vendors in this space are doing some really cool things, in addition to providing a lot of storage with a lot of performance at a reasonable cost:

  • Nimble.   Nimble’s hybrid approach includes compression in order to save some disk space and the company boasts more than 600 customers.
  • Tegile.  Tegile is a solution that provides hybrid capacity and performance while also adding support for Fiber Channel, NFS and iSCSI, making it an easy fit in almost any environment.  In addition, Tegile includes full inline deduplication to maximize the capacity in the box.
  • Tintri.  Tintri provides complete end-to-end performance monitoring in a vSphere environment. As an administrator, you always know exactly where bottlenecks are occurring at any location between the disk platter and the virtual machine.  This kind of insight in invaluable when it comes to troubleshooting.

These kinds of solutions are supplanting the likes of EMC and NetApp at the fraction of the prices of their more established brethren.  Customer satisfaction seems very high and the companies continue to report positive sales and services.  As startups, not all solutions are available everywhere in the world yet, though.   However, if you’re in the U.S., all three are available now.



  • ewilts

    I don’t know about EMC but NetApp has introduced hybrid aggregates to solve the same problem without giving up the rest of the traditional NetApp functionality. Whereas the startups are “working hard on deduplication technologies” and just recently added or are still working on things like snapshots, compression and replication, NetApp has had these features available for a long time. The reason the startups are still lower cost might just be because they’re feature-limited and as they add the features that mature storage vendors like EMC and NetApp already have, we may see the prices go up.

    I do like to see the things that startups offer (and we’ve got some non-big vendor storage in our datacenters as well), but don’t necessarily give up on your bigger vendors just yet.

  • rcommins

    Hi ewilts – I am from Tegile, one of the vendors mentioned in Scott’s post. There are several reasons new vendors like us can compete effectively against the big incumbents like NetApp. First is architectural advantage. We are purpose built for today’s storage needs. NetApp was purpose built, what, 15 years ago and was the disruptor then. Our in-line deduplication and compression is very different from NetApp. It is positioned very early in the data path, and yields performance advantages, as opposed to a penalty. This helps us reduce the amount of hardware required to achieve both performance and capacity requirements. The other is a business issue. It is very difficult for an incumbent vendor to make dramatic steps in cost reduction to their customer. It directly impacts revenue and gross margins. That is a big reason you see pricing models from incumbents that have such a high variable software component. They are priced to maintain gross margin integrity as the price of storage drops. All that said, I look forward to the day when wrestling with the challenges of being the big incumbent are our challenges at Tegile. I am glad to read that you have some non-big vendor stage on your datacenter floor. I encourage you to follow Scott’s advice and take a close look at these new disruptive storage solutions. There is some very exciting things happening these days.


  • I am from Starboard Storage Systems and we provide a unified hybrid solution and we do provide snapshots (Space efficient, high-performance, application consistent and consistency groups) and replication. We have succesfully replaced NetApp 3000 systems when customers have given us extensive criteria to meet. We are a lower cost for the overall solution because we have a better architecture that maximizes both the SSD and HDD resources and provide an all-inclusive software suite. Make no bones about it though. we have the enterprise features and we are not a slave to reservations, RAID groups, and stranded storage with HDD or SDD. http://www.starboardstorage.com . Don’t tar all startups with the same brush.

  • Pingback: Hybrid arrays provide CIOs with a balance between performance and capacity - CIOscape()