VNX = the worlds biggest, most bad a$$, swiss-army knife storage.

 

And by the title of the post, you know my blog posts aren’t ever approved by anyone :-)

Putting it into basic terms… VNX = 3x MORE EFFICIENT, 3x FASTER, 3x SIMPLER

What’s the VNX scoop in a nutshell?

• VNX was code-named “Culham”.   It’s a small village in the UK, but that small village also hosts one of the world’s leading fusion R&D centers that hosts The Joint European Torus.  It’s the largest and most powerful tokamak currently operating… Fusion + “Most Powerful” = VNX.  Hence the codename :-)
• VNX is EMC’s next-generation midrange storage platform – designed to do everything but deal with the kitchen sink, but in a way that’s simple, efficient, powerful, and affordable.
• VNX is a unified platform – every protocol under the sun  - Block and NAS (including MPFS and pNFS), but also Object storage.  While there is a block-only variant (VNX5100), this is mostly for specific OEM-type applications.   CLARiiON and Celerra aren’t end-of-life, but from now onwards, everything in the EMC midrange is unified.  Period.
• When we say 3x more performance, we’re not kidding.  
  • There’s distinct method to our madness.   Updates in EMC platforms generally tick-tock between “major software-only releases” and “hardware to fully leverage the new software”.   This helps us make sure that quality stays high, and we don’t change too much in one window.   SO – the innovations around FAST Cache, FAST, Compression, NAS Dedupe that came in FLARE 30 and DART 6.0 were built to leverage the fact that we new that the VNX would scale (at the upper end) to a platform that could have 120 Westmere cores and 240GB of memory servicing IO, and end to end PCIe v2.0 (more detail below).   Yup, you read that right.   120 Westmere cores.  Yikes.
  • Record-setting bandwdith and IOps in a midrange array.  We’ve demonstrated that in real-world workloads, sustained rates of 7GBps (ergo 50+ Gbps) or more.   For comparison, the previous generation (which was already good) was around 2GBps.   Do do this requires hardware (end to end PCIe 2.0, 6GB SAS backends, huge Ultraflex IO module density including 10GbE and FCoE).   BUT all that hardware, without the massive investments made in the software stack, there’s no way we would be able to use all that juice.  
  • And yes, in case you’re wondering, there are many benchmarks ready to go to prove it all – us vs. our own n-1 gen and us vs. the competition.
• People who claim VNX is a hardware rev without software changes, well… they don’t know what they are talking about :-)  VNX includes updates to software in a big way – some under the covers, much visible to the users.  Unisphere has been getting rave reviews (and is the tip of a big iceberg of innovation I talked about in the VNXe blog post), and it’s further improved in VNX (more detail below).   There’s other critical stuff in the heart of it too from a software standpoint.  As an example, in the previous software package, FAST Cache worked on block and NAS, but FAST was only supported on NAS.  This capability has been converged. 
• Software bundles just like with the VNXe simplify everthing.  Things like snapshots, replication and others have been simplified into 3 software packs.

In my personal opinion, if VNXe is the coolest thing we’re launching today, VNX is the most bad a$$.

Read on for more, including demos and details…

Hardware:

 

The mantra across the family is captured in 4 words.  Affordable.  Simple.  Efficient.  Powerful.  That mantra applies across the family, from the smallest to the largest.  “Powerful” becomes more and more important as you get bigger.

Whereas in the VNXe, we’re leveraging the C4 codebase to scale that same core functionality across cores in a tiny footprint, on the bigger platforms, it’s about scaling with more and more physical block brains and NAS brains – all managed via a single, simple Unisphere instance.

For the specs of the Storage Pool modules and the Filesystem Modules (and the platforms themselves)– see the table below.

Software

Across the VNXe and VNX family – the software has been simplified down to core software packs.

As an example, the total protection pack gets you incredible DVR like capabilities for primary storage - as many high-performance point in time copies as you want, and the best remote replication across file and block – all integrated with Unisphere for a single pane of glass for management.  Add in being able to see your data protection policy in a single beautiful app.  While we’re at it, let’s mix in application integration across all apps and VMware.  That’s all in that package.  Gotta love that.

Download the high-rez WMV here (BTW – product team, great demo – let’s not just post behind powerlink)

Download the high-rez WMV here (BTW – product team, great demo – let’s not just post behind powerlink)

Pause and think about the videos above.  Recoverpoint is already integrated with Unisphere.  Data Protection Advisor will be soon.   Think about the C4 implications in the VNXe post I made earlier.  Get it?

Other tidbits (there’s so much stuff in this launch, awesomesauce like the below are “tidbits”)

• FAST is now supported on NAS volumes
• Hardware carousels now fully integrate all NAS and block elements
• ALUA performance improvements through the roof
• Atmos integrated into Unisphere
• Improvements in the Recoverpoint integration with Unipshere
• Isilon link and launch (that was fast – kudos to the engineering teams!).   Highlights how fast we’re getting re: Unisphere.   C4 does the same thing not just for management/look&feel – but for core stacks…

Just awesome stuff – I think our customers will love it… What do you think?

Comments

[Andrew Fidel]

One thing I've never understood is why you can't put more cache ram in the mid level offering. You're using industry standard servers and ram is cheap so why can't I get an array where 99% of my read requests are serviced from cache instead of 95% (just an example) without paying a small fortune to move up to the bigger heads?

[vishal kumar]

Hi Chad,

So what is VNX running on? FLARE, DART or CenteraStar ? Could you also explain more about how it can scale. i.e. how many SPs/DataMovers/Centera Storage / server nodes?

Thanks!

[Riker82]

Hi Chad,
I've been invited as customer @ EMC Italy in order to view the EMC Roadmap some months ago. I was advised about this new generation "Clariion" and they told me that FC disks will not be available anymore on our mid range storage.... I was wondering that was a joke... How EMC can ask to their customers to throw away all the existing DAE and disks (in my case more of 13x DAE3P/4P) and asks to buy new SAS splindles without re-using our old disks (I've already did in the past when I moved from CX3 to CX4)?

How much SAS will cost? In Italy I've just bought 600 GB FC 4 Gbps for 1400/1600 € per disk!

[Mike Dunn]

Chad,
Where could I find which exact (not just intel 5600) model proc is in each model of the VNX family?

Thanks,

Mike Dunn

[j]

VNX is running FLARE...

[Andriven]

@Andrew Fidel -- this is the whole point of FAST Cache (or at least most of the point). While you can't increase the main array cache, you can augment it to pretty massive degree with FAST Cache.

[PowerPath]

I hope new DART works as active/standby mode within two data movers. So we can distribute works loads on two data movers, just like we spread LUNs on two SP of VNX. It's really good for small setup with only two data movers.

And I really hope DART has scale capability, scale up, and NAS cluster, but not sure if it still fits the marketing strategy.