Recommendations For All-Flash Storage Arrays [Updated]; Aiming Beyond Simply IOPS and Low Latency

[Updated: I’ve added a new recommendation and some new comments on the flash storage landscape.]

A lot has happened in the past year in the flash storage market. I think most flash storage array companies or the flash storage business units within large storage companies should be growing at least between 10% to 20%. The market is sufficiently hot that it has attracted a lot of customers and a fiercely competitive crowd of flash storage array companies.  Today’s post is about what I would recommend if a friend asked me what flash storage array I would recommend and depending on what they were doing. These three companies offer solid offerings with solid and very high performance with excellent features and all three companies are managed extremely well. All three are growing at hyper-growth rates.  In the past I have offered Key Questions To Ask Your Flash Storage Vendor Before You Buy.  More recently I have highlighted scalability and aggregation of the storage as two key features in the post, Why Lego Makes Sense in Toys, Software, Servers and Storage. Increasingly, smaller, more nimble storage companies are coming out with excellent flash storage solutions.

If you are looking for flash storage – the reality is that almost all the all-flash array vendors are fast enough, in fact the fastest of the lot are actually the weakest in terms of storage/software features and that should be an important aspect of your focus.  Companies slog along trying to retrofit various solutions into their architectures – basically bolting on features.  Take a storage feature such as dedup .  I’ve written before about how some vendors , who until recently didn’t have dedup, have written some very defensive posts against dedup. Today these vendors have been forced to recant and adopt dedup.   If you don’t get the storage features you need – you can’t use them. And if you get them the wrong way – they cost more. It’s worth paying attention to the feature set in the array you intend to purchase and make sure it has what you need.

Recently, someone asked me what flash array I would recommend to a friend. Here are my top three all-flash storage companies to look at. No slight to other companies producing arrays, they all have good aspects but these three all-flash arrays excel in providing both the technology (both in hardware and software), product innovation and product stability I would want in a recommendation.  Two are scale-up arrays and the other a scale-out array. These three all-flash arrays provide excellent performance, a wealth of features, consistency and best practices. In addition, they offer support for various virtualization environments such VMware, Citrix, OpenStack and Cloudstack. These arrays have a large number of companies using their solutions successfully in a number of different settings.  If you are looking at an all-flash array – keep in mind questions you should be asking – see : Top Thirteen Questions to Ask Your (All) Flash Storage Array Vendor. Here are the three all-flash arrays I would recommend to someone   – keep in mind that these three choices transcends speed – these arrays will provide high IOPS and low latencies – they offer a strong wealth of useful features. I have added their strengths as a side chart. One of these companies has the advantage of offering also hybrid arrays.

  • Solidfire (now part NetApp) offers an excellent scale-out architecture that is without doubt ahead of all the other vendors when it comes to providing cloud storage features for cloud storage providers and entesfhigh2rprise cloud deployments.  They have won over a large number of cloud providers.  It is not a surprise – they have built in a number of critical scale-out and storage features which I have previously reviewed.  Solidfire scale-out arrays scale up past 100 nodes to provide a highly available view of storage with quality-of-service controls and all the usual suspects of data reduction built in to the operating system (including replication, dedup, compression, etc).  In my view, focusing only on storage misses an important point – storage lives within larger ecosystems.   Solidfire’s work on OpenStack, Cloudstack, Citrix and VMware frameworks offers a solid well-rounded group of storage companies with a focus on the complete cloud solution. You can look at the post OpenStack Announcement: Solidfire/Dell/Red Hat Unleash Solidfire Agile Infrastructure (Flash-Storage-Based) Clould Reference Architecture to understand their involvement in cloud solutions. The Solidfire architecture allows from four arrays to 100 to be clustered and provide petabytes of flash storage that are highly available. The can make use of either iSCSI or 8/16Gb Fiber Channel.  Their website is http://solidfire.com
  • Pure Storage has demonstrated that a company can create an increasing larger customer base by bringing to market a solid scale-up array with key features that purehighcustomers doing databases and virtualization deployments will find of value. Pure Storage has done an excellent job of showing why data reduction is important.  You can read one use-case Data Points : Delphix & Pure Storage Database Benchmark Shows 10x Price/Performance Gain.  This particular use-case shows Pure Storage’s strength in data reduction as it applies to databases.  There are a number of other use-cases where data reduction serves as a powerful feature – virtualization and specifically VDI makes particularly good use of data reduction.  From the post, Accelerating VDI with Flash Storage, you can see some of the benefits from data reduction and snapshots.  One of the particularly useful visuals on their web sites is the FlashReduce Ticker which is a scrolling ticker that indicates the average data reduction and average total reduction benefits seen by their customers. Their website is http://www.purestorage.com
  • Tegile is a relatively new company that offers both hybrid (disk + flash) and pure flash arrays. This offers distinct advantages over pure flash array companies – they can offer both pure flash ( extreme high performance) or disk + flash (high capacitytegileFA with high performance). Both have huge advantages over disk-based arrays. The speed at which Tegile is gaining tractions should cause concern to their competitors. In addition, the ability to offer their customers a choice is significant advantage. For example, Tegile’s hybrid arrays delivered between 25-40x the performance of their disk-based arrays at BYU Hawaii. Other customers like Saudi Petroleum are standardizing on Tegile.  Meanwhile their new all-flash arrays have already gained significant traction. A recent example, Oklahoma Hospital supports a large VDI deployment and they recently moved to all-flash arrays from Tegile.  Support for data reduction features is a big win for VDI deployments and one of the many data reduction features Tegile offers is deduplication and compression. The advantage of Tegile’s product offerings provide both a way to offer high capacity and high performance coupled to advanced storage features. Tegile’s offering is built on an excellent storage platform – ZFS.  Their website is at http://tegile.com.

Of course, there are many more all-flash arrays out there, many are also very good.  In my opinion, if you are looking at all-flash arrays – Solidfire and Pure Storage have stood out and Tegile has emerged to offer a highly competitive offering.  It’s not an accident that in Gartner’s latest All-Flash array critical capabilities survey Solidfire and Pure Storage out-performed the rest of the field (with the exception of Kaminario which is also a very good offering) they have a wealth of features. You can read more about this in the post Gartner Releases Flash Array Critical Capabilities Study – Solidfire, Pure Storage Come In First. Increasingly these companies are causing fits to the older flash storage companies that have been either trying to regain traction.  If you don’t think this is causing fits all you have to do is look at IDC’s Worldwide All-Flash Array and Hybrid Flash Array 2014-2018 1H14 Vendor Shares. Just two (Solidfire and Pure Storage) accounted for over 18% of the flash capacity shipped. And both these companies have been growing at 700% (Solidfire in 2013 and was growing at 50% growth quarter over quarter in 2014..meanwhile Pure Storage also came in at 700% in 2013 ). Tegile just recently released their all-flash arrays – they have been able to watch and avoid many of the mistakes of their competitors.  EMC comes out looking good in this particular report. If you look at the Gartner Critical Capabilities Report you can see how these smaller companies are faring competitively – remarkably they are ahead of the pack.  It will be interesting to see how not only Violin Memory but also NetApp staves off these newcomers – let alone battle against large companies like IBM and EMC. It is a very rough competitive landscape.  Regardless, this trio of Pure Storage, Solidfire and newcomer, Tegile, are having an a huge effect on the flash storage landscape.

Agree ? Don’t agree ?  Feel free to send me a question @ cld9731@gmail.com.

 


gotostorageGo to more posts on storage and flash storage at http://digitalcld.com/cld/category/storage.


Seriously. Sous-vide Cooking with Clojure Transducers

Seriously.  Someone has taken the trouble to manage sous vide cooking using Clojure. In what some might describe as a science experiment there is a nice write-up on linking a Clojure cooking app to a sous vide cooking environment.  You might ask why is this interesting ? Primarily, because in the ‘internet of things’ more and more processes are interacting with remote, sophisticated software.  In this case, the use of Clojure Transducers as part of the software that manages cooking is an example of where things are going and how software could and is being developed to control things.

cljcooking01

 

Fusion IO Blows Through 1 million IOPS to Nine Million IOPS in Transaction Logging App Use-Case

Everywhere there is this need for speed. One flash array vendor thought that benchmarking a single VM to achieve a million IOPS was particularly useful or interesting. Marginally.  However, I found that the real example of speed comes from Fusion IO and it has demonstrated it could deliver 9 million IOPS from a single 365 GB MLC Fusion ioDrive2 card. Watch the video the number is actually 9.6 million IOPS.  Unbelievable.

fioshowcase

And a quick brief on the use-case :

fio9milltx

Here is a nice detailed presentation/talk from youtube on it :

And a presentation on it :

fio9millpreso

The key aspect beyond the sheer hardware effort is the work done in building programming interfaces via their SDK.  By focusing not simply on sheer hardware but also on building useful programming interfaces for software developers, Fusion IO provides the tools to achieve very high IOPS.  Many flash companies have focused on hardware – but here Fusion IO shows that a focus on hardware and software and file systems is at the heart of their tremendous performance.

 

More on Clojure and Storm, a Distributed, Real-Time Computation System

In yesterday’s post I brought up Storm. There is a nice presentation on Storm from Nathan Marz of Twitter.  He explains the workings of Storm, a distributed fault-tolerant and real-time computational system currently used by Twitter to keep statistics on user clicks for every URL and domain.

 

Note that that there are some examples of Storm topologies.

storm-starter

 

Here is some details on how Twitter was using Storm in 2011.