It’s been a while since the VNX2 was born: September 2013, I remember it very well. Being a part of the EMC Elect, I was invited to be at the actual launch in Milan (Italy) and what a ride it was! The whole launch was wrapped around Formula 1 technology and it sure was “speed 2 lead“. That “old” VNX2, which I’m still perfectly happy with by the way, was a revolution in my humble opinion: multi-core everything, in short MCx. And yes, it was like everything just went faster, smoother and better.
But with new technologies popping up every so many months now, it was time for a new mid-range storage array. Flash storage isn’t a novelty anymore, it’s a must! And the “old” hybrid arrays were fine, but needed some fine-tuning. With flash devices growing bigger every quarter or half a year and faster as well, the whole back-end needed an upgrade. The old 6 Gb back-end (x4) needed an upgrade.
A long time ago, there were two array families called VNXe and (the more scalable) VNX. These two were meant for two different types of customers: small businesses and medium to large businesses. With the Unity line-up, it’s now one new big happy family.
And one feature already catches our eye: there’s two types of Unity models: The F (flash) and the hybrid. And you can actually tell them apart by looking at them: the All Flash Array has a silver logo and the hybrid array a combines blue/silver logo.
Oh and there’s the virtual storage appliance as well (VSA), but that one has been around since EMC World last year already. Sure, there were improvements, but it’s a grown up virtual storage appliance now and you can even get support on it as well!
Under the hood it’s a whole new operating system as well: no more hardened Windows anymore, but it’s Linux! And what’s more: the need for Java is gone! That last part will make a lot of people very happy. No more dependencies on particular Java version and incompatibility issues. It’s now html5! YES! Do I really need to explain the benefit of getting rid of Java? Nope!
Also, the Unity finally has native vVOL support for VMware. This will give more control to storage tiering to the VMware front-end.
Yes, the first listed appliance is the virtual one! And it’s supported! Up to 4TB for free even. This makes the VSA perfect for small OTA environments. If you need larger virtual configurations you can get support on the VSA in these steps: 10 TB, 25 TB and 50 TB. Bigger environments are suitable for the physical Unity machines.
The VNX 7600 and 8000 series will remain to be the best choice for enterprise solutions up to a million IOps and 6 PB of storage.
There will be three DAE types available: a 12-drive, a 15-drive and a 25-drive shelf. The 12- and 15-drive models are for the 2, 4 and 6 TB 7200 RPM drives.
The supported drives at GA will be the 600 GB 15 k as well as a series of 10 k models: 600 GB, 1.2 TB and 1.8 TB. Also note that the old 6 Gb DAEs are not supported on Unity systems and all drives for the Unity will now have the new 4 kB bytes per sector format (4160 bytes). Only the SSD devices will continu to use the 520 BPS format.
Which features are available?
The VSA is somewhat limited (no FAST Cache), and the AFA with the absence of storage tiering and FAST Cache, but that’s quite obvious. Having said that and with the availability of 3 flash tiers, not being able to move data between the 3 flash tiers seems a bit odd.
CloudIQ offers powerful support capabilities to set smart alerts and quickly resolve problems. A new cloud-based management portal will be introduced that lets you quickly get health scores and easily monitor and manage your environment. This could mean that EMC will gather (anonymously) intelligence about your environment to see how it is used. Based on different metrics, future upgrades can make sure you will get what you need, even before you thought you needed it. But wait!
To share or not to share
What if a customer doesn’t want to share these usage statistics? Everyone will be free to share or not to share this anonymous data. It’s obvious that cloud based usage gathering will create a big data alike intelligence that will give EMC the benefit of predicting where customers are headed to, but even if companies choose to opt-out, there will be enough data available to successfully develop firmware updates.
VASA, vVOL’s, SMB3, NFS, CIFS, and more capabilities will be available from the start. From one interface, protection schedules for file, block, and vVOLs can be created using unified replication and snapshots. Up to 80 TB in a single 2 U shelf will be available from the start with larger capacities on the way.
The Mad*Pow test
EMC had Mad*Pow test the new user experience and some promising findings were:
- “Easy setup, intuitive navigation, and robust visualizations provide a very easy interface to manage storage arrays”
- “EMC Unity provides easy navigation, a clean look & feel, and intuitive visualizations that help with storage array management”
- “Intuitive, facilitated by a clean, modern, and straightforward look & feel”
Other pieces from the research team:
- “A Modern, Clean Look and Feel”
- “visualizations that truly help users manage their storage arrays”
- “Easy setup, intuitive navigation, and robust visualizations”
- “You don’t need to be a storage guru to use these systems”
- “EMC is setting the standard for storage simplicity”
No more physical datamovers and control stations
With the even faster CPUs and large memory sizes, it became possible to virtualize the physical file hardware (datamovers and control stations). The Celerra legacy has finally left the room. The Unity is a mature looking storage array with file and block capabilities built-in one box, instead of multiple combined boxes.
Graphical User Interface
The GUI looks familiar: Unisphere Central sure has had some influence. It looks slick and invites you to start drilling down for information. It looks like managers can now take a look and get the information they need, instead of storage admins that needed to do magic with spreadsheets and graphs they had to create themselves. Yeah, the GUI looks very 2016!
The secure remote services are now based on docker and integrated inside each Unity machine. The VNXe and VNX1 and 2 had that same functionality since about two years as well and yes, a dedicated ESRS appliance is also supported. An ESRS appliance has its advantages when a customer’s environment consists of more than just a few EMC supported machines. With ESRS, a (centralized) secure way of granting access to EMC engineers to support an end-user environment is a huge relief. No more web-ex, no more teamviewer. Just let EMC do its job and don’t worry about having to spend time on granting access to EMC’s engineers.
What’s in the box?
In the last few years there have been quite a few new developments in the development of flash. SLC, MLC, eMLC, TLC, 3D TLC, 3D Xpoint, MCS. It’s so much that flash isn’t simply flash anymore. Each kind of flash has its own characteristics and vulnerability to wear. Some brands have even embedded wear leveling mechanisms inside the flash device of wear prediction mechanisms, so wear leveling doesn’t even have to take place. In the rotating disk corner, I present you: slower NL-SAS and SAS drives (Fibre Channel has left the building) and in the solid state corner the most popular flash devices I actually just mentioned. But wait: If NL-SAS and SAS are two tiers that differ in performance, does that mean there are multiple tiers in the flash corner as well? Yes sir! Some devices are meant to be written to extremely often and they need to have a good endurance as well. SLC used to be the primary type for that kind of IO, but there is no more SLC flash (probably called flash-1). But what if data just needs to be accessed quickly, but just not that often? Vendors came up with new types of flash that can be described as “10 writes per day, 3 writes per day” or even “1 write per day” (later this year).
You now will have the choice to use the traditional eMLC flash devices for FAST Cache as well as for FAST VP (previously only for FAST VP). The 2nd flash tier is now only for a single tier pool. Let’s call these two flash tiers flash 2 and 3:
- FAST Cache: flash-2 (only the 200, 400GB and 800GB (800GB only in Unity 600) flash devices)
- FAST VP: flash-2 (including the 1.6TB device) in a multi tier pool (with/without SAS and/or NL-SAS)
- and flash-3 (only in a flash-only pool, so no multi-tier pool); available with 400, 800, 1600 and 3200GB flash devices
So in short FAST Cache will use fast-2 and in a FAST-VP multi-tier pool you can use flash-2 as well. Flash-3 is for a flash-only pool only!
Built on Suse Linux Enterprise Server (SLES), the Unity is now more flexible than ever.
The new Unity operating environment is now modular with room for expansion and atomic improvement. Wow, I like the word atomic 🙂
With server-based storage, purpose-built storage, converged infrastructure and even cloud based storage, the whole playground needs a layer on top of all this: ViPR to rule them all. Eventually SDS is where we’re headed anyway. It’s all about building blocks and when the underlying blocks are industry standard, almost atomic, they’re easier to bind and manage. And with 1 storage family underneath, the Unity, we’re headed in the right direction.
The VNXe 3200
A new file system will be available on this next Gen VNX variant. The new file system will be 64 TB, offer space shrink, pointer based snaps, replication, quotas and a host of other features pertaining to virtualization and enterprise. A phase 1 version of the file system – minus snaps, replication and a few other features – is available on the VNXe3200 today, formerly known as the KittyHawk. It’s no surprise that the Unity as we see it today, had its roots in the KittyHawk, that started back in 2013 already, right after the VNX2 was born. I’ve seen the VNXe 3200 grow with my own eyes, being part of the beta KH program and I was impressed by its capabilities already. It’s no secret that I’m a block guy, I don’t care too much for file oriented storage, but this growing and shrinking of filesystems sure got my attention!
And with flash optimized, new algorithms had to be developed to make optimal use of flash and reduce wear.
The operating environment has become flash aware, so to speak. It’s no longer an adjusted OS that was based on rotating / slow disks, but a modern intelligent environment that makes optimal use of the latest technology in the industry.
The unified pool has now become intelligent: file-systems can now grow and shrink as needed.
And pools are now true unified: whether storage is block accessible or by using a file-system, it’s all based in the same pool-structure. There’s no need to have this separated anymore.
Clones and snapshots
Up to 256 clones and snaps are now possible per object and all writable! Everything is now pointer based and there’s no need to make reservations for snaps anymore: everything is now pool and pointer based. And snaps from snaps goes up to 10 levels deep if you need to have that.
The magic word is “simplicity”
Installation is said to be child’s play. A new Unity should be racked and stacked in minutes, well, I need to see that with my own eyes, but it’s a bold statement, so there must be some truth in it!
Management has become intuitive, just like Unisphere Central.
And with the cloud based monitoring and proactive assist, troubleshooting should now become a back-ground task for an EMC engineer while the storage admin can perform other tasks.
To sum things up:
Well, that’s looking good. No more Java, a nice management integration, an intuitive management interface and cloud based analytics (with the option to opt-out if data privacy is a concern) and more to come. The future is today!
And NO MORE JAVA!!!
Reduce Total Cost of Ownership by
- Minimizing time and effort to implement
- Increasing self-service
- Making intelligent, just-in-time recommendations on best practices and risk management
Deliver higher uptime, increased performance and easier capacity planning through
- Analytics and increased proactive monitoring
- Cross-customer intelligence
Provide ubiquitous access to information and tools
With CloudIQ customers can profit from big data analytics:
- Enable Cross-Customer intelligence and comparisons
- Best practices for Pools – capacity comparisons
- User-defined threshold for % of Install Base
- Notification on device: we’ve noticed that other customers with xyz similar config benefitted by doing 123; would you like for us to proceed with that now (i.e. firmware update on drive)
That’s real proactive support.