Today, EMC announced VSPEX Blue, which is an awesome new Hyper-Converged appliance - designed to deliver great and unique capabilities, global supply chain and support - and do it in a small, inexpensive package. VSPEX Blue delivers everything you need for a broad set of workloads and scales - and ABOVE ALL - it does that with simplicity.
Appliances tend to be much more fixed configuration (to some degree in both the dimensions of the software and the hardware) than the engineered system variety of converged infrastructure (think Vblock/HP ConvergedSystem) which in turn have less variety than the reference architecture variety of CI.
VSPEX Blue is an appliance. Simple. Turnkey. Scales from tens to hundreds of VMs with ease.Read on to learn about all the goodies in VSPEX Blue!
Why the name "VSPEX Blue"? Well - VSPEX is a wildly successful channel program (~$2B of channel-led revenue!), and this is appliance has a channel-led go-to-market. But, unlike VSPEX in general, this was a case where the support and experience was a "full product", not a reference architecture. The support is provided end-to-end by EMC, and the EMC logo is blue hence "Blue" :-)
So - what have we been working on since the public unveiling of the EVO:RAIL program (one of the foundational ingredients that makes up VSPEX Blue) in late 2014?
VMware's EVO:RAIL program enabled OEMs to create hyper-converged appliances with relative ease: it created a simple software package (including an "all in" licensing model for all the VMware software), a simple management tool (EVO:RAIL Manager), and a proscriptive (quite proscriptive) hardware configuration for every OEM. There is no "EVO:RAIL product". There are EVO:RAIL OEMs - and EMC is one of them.
We wanted to enhance the EVO:RAIL experience - but knew that if we did it in a way that "broke" the appliance experience, that somehow degraded the extreme simplicity - then we would have made a "franken-appliance".
The first innovation was to directly extend the VSPEX Blue management model - the experience, look and feel are directly integrated with the EVO:RAIL Manager itself.
The second innovation was to think about the end-to-end support model of an appliance. Surely, SURELY people would want deep hardware failure detection (including low-level BMC info), not just the stuff that surfaces up to ESX/vCenter? And of COURSE, surely, SURELY people would want that completely integrated with the vendor providing support for the whole appliance (in this case EMC), right?
VSPEX Blue Manager takes physical information from the low level hardware (customization in the BMC) and also aggregates it with the embedded info gathered in the instance of VMware Log Insight that is included in all EVO:RAIL based appliances.
The physical appliance state, along with low-level fault information (not in EVO:RAIL on it's own) are shown in the extended VSPEX Manager UI - all without leaving the appliance experience.
This information, along with other system state information (but of course, not user data!) is also constantly sent to EMC because VSPEX Blue integrates with the EMC Secure Remote Services (ESRS) capability. This means the customer has proactive monitoring, and it also fully integrates with http://support.emc.com tools. These two things mean something very important:
- Low level hardware alerting and proactive monitoring improves overall system availability rates by 15%. That's a LOT.
- Rapid remote diagnosis and support shows up as a 5x faster time to resolve issues.
This is data that we have across hundreds of thousands of customers that are using the ESRS platform (or not) across the entire EMC customer base. It would absolutely suck to have an appliance that doesn't do this fundamental thing one expects of appliances.
Cool. Check it out (along with the whole marketplace) below!
The third innovation was to think about additional value-added capabilities that customers would want - and how to make these value-added capabilities something that was extensible. In VSPEX Blue - there are 3 additional capabilities out of the box: 1) additional NAS/block storage from public clouds; 2) Enhanced backup capabilities; 3) Enhanced local and remote replication capabilities.
First thing - the VSPEX Blue Marketplace makes adding capabilities as simple as everything in VSPEX Blue - all in the VSPEX Blue Manager, with the consistent look and feel.
So - let's learn a little more about the 3 additional capabilities we added:
- The embedded Cloud Array Virtual Edition solves one of the challenges of the Hyper-Converged appliance models: relatively meager storage capacity.
- With these dense 2U x 4N (2U form factor, 4 compute nodes), you only get 4 x 2.5" drive slots (for 16 in total). For comparison, on a VNXe (a storage appliance that also targets the lower end market), you can get up to 25 x 2.5" drives in a 2U form factor. In the bigger bad boys in storage world (think VNX/VMAX type things) - we can get 120 x 2.5" drives in a 3U form factor!
- Further, with the software storage stacks in hyper-converged appliances (in this case, VSAN, but also applies to other approaches) - inevitably there is a need to protect from total node failure by distributing copies, so storage utilization (pre data reduction techniques) is always going to be at least 50% lower than the raw capacity
- Now, the local capacity may be enough for the VMs on VSPEX Blue itself - but what about about other things you might need if VSPEX Blue is your "datacenter appliance"? Bolting on a completely separate array would be possible, but seems to break the "self-contained appliance" model. We needed to do something different.
- .So we did! The included Cloud Array VE is a embedded software-only solution that lets you use external public cloud object storage (AWS S3, vCloud Air, etc - more than 20!), and it can cache the content locally - presenting it ultimately as NAS (SMB/NFS) or iSCSI targets. Cool! This is so awesome - check it out below:
- VSPEX Blue includes a 10TB storage and 1TB cache license with the VSPEX Blue package. That means that if every appliance includes 12TB of physical capacity, the NET what a customer gets as a total is 22TB! Further, this included license is per appliance, so if you scale the appliance from 1-4 appliances - each adds 10TB of additional cloud-based storage. And hey, if you want to license more - you can.
- No matter how many VMs you have - backup matters.
-
You can, of course, use VDP for backup (or other backup tools), but backing up the appliance itself has the classic "backing something up onto itself can be. problematic" issue.
- So. We integrated full VDP-Advanced (VDPA) capabilities into VSPEX Blue! EVO:RAIL always includes a vSphere Enterprise Plus license, but VDPA by default isn't obvious. By surfacing it, this means customers can easily attach a Data Domain appliance to backup and recover their VM images.
-
- No matter how many VMs you have - remote replication matters.
- Recoverpoint for VMs provides awesome per-VM replication capabilities (and rich data services like consistency groups) far above and beyond native vSphere Replication.
- This was announced at VMworld 2014, and you can see demos here. Recoverpoint for VMs can also be used separately, and is available for simple & frictionless access here. But, of course, people would want it to be part of the appliance.
- So. We included 15 free full Recoverpoint for VM licenses per appliance (if you have a cluster of 4, that would be 60 VMs) that can be replicated with full blown support with VSPEX Blue.
What about the hardware? Who makes it? Who supports it? The answer is "EMC".
While people don't think of the EMC brand as a server vendor (and we aren't - you can't buy a server from us), we have been a volume shipper of Intel x86 hardware for eons in appliances. VSPEX Blue uses the "Phoenix" code-named server platform (which is our "dense balanced storage/memory/compute" platform). For example, we have been using Phoenix modular systems in our Elastic Cloud Storage appliance (in addition to VERY dense capacity storage enclosures). We're not a one trick pony when it comes to this domain - there are "Python", "Rinjin", "Dragon" codenamed servers that each have a target across the broad set of use-cases (stay tuned).
In the case of these appliances - the hardware is important (a little bit of extra pride and quality goes a long way), but not THAT important. People will perhaps read a little too much into the "EMC is in the SERVER BUSINESS!!!" point. We aren't really. We build solutions that happen to need servers. If we find that the solution is best served via an OEM style model (for example, Cisco UCS blades in Vblock), then we OEM. If we don't find an OEM that is right, we build it.
This is what the hardware looks like minus the faceplate. BTW - the team (Chad Dunn, one of a small number of my EMC "Chad" brothers!) recorded a tear-down (that includes some hints at ideas we have), also below:
It's also important to know that VSPEX Blue has a really, really unique channel business model (which sometimes is as important as the technology). This is less meaningful to to the end customer, but important for the full value chain.
- First of all - while the EMC field has goals for VSPEX Blue - it is a channel only product. Yup - it's not even in DXP. This is great news for EMC partners. You have a great product (having done 50+ NDA customer briefings over the last month, they LOVE VSPEX Blue!) - but you also have the full strength of the EMC field at your side, but they cannot deliver the value to the customer without you.
- Second - we have designed the whole value chain to make it a great business for the customer, and for the distributor and partner. This is an interesting sidebar dialog. Ping me if you want to understand more.
Like all things - I'm sure that there's lots to improve on, and I'd love to hear your thoughts!
Hint: 2015 will see a TON of action from EMC when it comes to Hyper-Converged architectures, including a lot more on VSPEX Blue, but also other things as well (at broad scales, and for all sorts of different workloads). EMC is currently the market leader in the broad converged infrastructure space with Vblock, and will also double down on Vblock architectural models. We think that the CI market is best served with a portfolio. VSPEX Blue marks our first offer with a hyper-converged architecture - and in our CI portfolio is the offer that targets simplicity above all else and delivering the ultimate appliance experience. HANG ON TO YOUR HATS - 2015 is going to be awesome!
more than 20 external public cloud object storage: where can I find a list of these public cloud providers?
Posted by: Lieven Pevenage | February 03, 2015 at 01:00 PM
Hi Chad,
I really like the idea of EVO:RAIL and have been looking into it for some time.
Great to see that EMC is now on-board and adding some additional value.
I do have a couple of concerns that hopefully you can help with:
1. Why are the licences tied to the hardware?
2. Why are the licences Ent+?
3. Why can we not upgrade one node at a time?
4. Why can we not have different disk specifications/upgrade them?
I did cover this all in a blog in more detail at http://blog.snsltd.co.uk/vmware-evorail-or-vsan-which-makes-the-most-sense/
Your thoughts would be much appreciated.
Many thanks
Mark
Posted by: Mark Burgess | February 05, 2015 at 11:11 AM
Cloud Array VE included is awesome - I also loved the "raw" Cloud Array demo with the "oops ... let's retake that" sections included :D
Posted by: Pete_geek | February 11, 2015 at 07:11 PM
Hey Lieven,
I'm not sure if EMC will retain the same public cloud providers on CloudArray (I can't imagine why they wouldn't), but the list of partners for CloudArray (back when it was owned by TwinStrata) can be found at the following link: http://www.twinstrata.com/cloud-storage-providers/
Posted by: Nick O. | February 19, 2015 at 08:47 PM