NetApp AFF (All Flash FAS) Review
Its ability to handle the load which we throw at it

How has it helped my organization?

Ease of use: We're familiar with the NetApp platform and ONTAP. We're comfortable with the tool sets that it has. We've been trained on it as a group for quite sometime. We started out with IBM-branded NetApp with 7-mode. We've grown from 7-mode all the way into ONTAP 9.0. The cross training amongst players or team members allows us to help each other with issues that we deal with on a regular basis. We find that there's a lot of value in that.

We use it for a storage location for Riverbed centralized storage. We use it for VMware, VMFS volumes, and for our VMware platform. We also use it for iSCSI and for regular RDM server storage. We use it primarily for block-related storage.

We use it for multiple apps. It's enterprise-wide. We have eMARs. We have what they call the Obamacare Exchange running on it, and HBE for the State of Kentucky. We have a lot of VMware running on it, which have 1000s of servers that their VMDK files are nested in VMFS volumes which run on the AFF8080.

One of the primary reasons that we went with the AFF was because of the dedupe, the compression, and that it's not software-based, but it's hardware-based. It's inline.

Learn about the benefits of NVMe, NVME-oF and SCM. Read New Frontiers in Solid-State Storage.

What is most valuable?

  • Performance
  • Integration into the ONTAP
  • The cost of the product itself

With the compression and dedupe, it's not necessarily a one-to-one gigabyte for gigabyte, where the compression and the dedupe allow you to buy a lot less, but to obtain a lot more storage capacity at the same time, hence getting the performance of SSD but they are not impacted by the two components of dedupe and compression. In summary, they don't get in the way of the performance of the product.

What needs improvement?

I would like to see a little more integration with some of the core fundamental components of OCI as part of the ONTAP OnCommand Manager, instead of it just being either all OnCommand Unified Manager or being able to see OCI and all of that it does. With what we pay for a node-pair and the OnCommand Unified manager, there ought to be at least a third of that integration in performance monitoring and alerting, and there's a lot there, don't get me wrong. We've got all the alerting and everything, but there should be a little more of the OCI bundled into the OnCommand Unified Manager.

In future versions, since we own every license that NetApp has except SnapLock. I would like to see SnapLock integrated into the platform, and not be an additional cost for a license.

We had every license when we purchased our platform. We're a major player in NetApp when you consider our total platform, as far as all the data that we manage is around about 12.5 to 13 petabytes. When you consider the size of our investment into NetApp, whether it's the AltaVault storage grid, E-Series 2800, FAS8060, 8080, or the AFF8700, we have a substantial purchase into all of their products at both the Commonwealth datacenter and also the alternate datacenter. When you consider we own every license that they have, except the SnapLock, and that's the one that we need the most right now for our stakeholders' legal purposes.

What do I think about the stability of the solution?

It's pretty good overall. With the auto-supports and the support SEs which are on staff when stuff goes bad and we have bad hard drives, we found that it's a pretty stable platform.

Also, all storage platforms have issues. There's things that go wrong with all storage platforms. There's no magic platform out there, but the response of the NetApp support staff, engineering, the ticketing, and the people whom help when you call in a ticket, they're very responsive and that also has a great value.

What do I think about the scalability of the solution?

It's very scalable. Right now, at our primary site, we have four FAS8060 nodes. We have two node-pairs of 8080, and we're adding an additional node-pair of 8080 along with a node-pair of A700. At an alternate data site, we've got a node-pair of 8060 and a node-pair of 8080. We're adding a node-pair of 80200. For the upgrade at the primary site, the only portion of that would be considered risky is it has to go through change control when replacing the intercluster switch. Because we're expanding beyond the capacity of the original switch that we purchased, and it's very scalable, and we like the product.

How is customer service and technical support?

They're always very good. Whether I contact them online or whether I call in, they're very diligent in following up and making sure issues have been resolved before they close the ticket.

Learn about the benefits of NVMe, NVME-oF and SCM. Read New Frontiers in Solid-State Storage.

Which solutions did we use previously?

We have multiple platforms. We have EMC, VNX7600s, and we just got rid of a VNX5600 and 5400 that were not able to keep up with the compute for what we were driving through them. We had on one of those systems, the VNX5600, we had 250 terabytes of free space that couldn't be utilized because the processing power on the platform couldn't keep up with what we needed. It was over-utilized, therefore we went with NetApp because it has the ability to handle the load that we throw at it.

How was the initial setup?

I was involved in the initial setup. It was somewhat complex, because we did cutover from 7-mode, where we stood up a brand new platform, were having to move the data from one to the other, and were dealing with the outages that were involved, but also going from the seven-mode to the ONTAP and the clustering and how it is different.

I also do a lot of the infrastructure, as far as the fabric management, the ports, the trunks, and the fiber-connections from the NetApp platform or the NetApp cluster to the IBM Brocade Branded Directors. I do all of it: the zoning and the fabric management. It's very detailed and very complex. You have to really know what you're doing in order to get that set up properly. It is not on NetApp. That's just in general. If it was any system, you would have that to deal with.

Which other solutions did I evaluate?

Every time we go through an upgrade process or we have a new purchase, we look at what functionality is offered by each vendor/manufacturer and we don't purchase based on fidelity to a single vendor. It has to be based on:

  1. Monetarily does it make sense for us to go with that vendor. Are they willing to work with us on the price?
  2. What they're offering. Does it give us what we need?
  3. Does it allow us scalability in what we're doing?

We just got finished purchasing a new node-pair of 8080, AFF8700, and an 8200. If Unity would have come in at a comparable price, we could have gone with them. We didn't simply because of the scalability of the product.

What other advice do I have?

Look for these three major components when researching a similar product:

  1. Supportability with tech support
  2. Scalability
  3. The stability of the platforms.

As far as AFF, we've had far better response and longevity of the actual drives themselves because they don't wear out as fast as a spindle drive does. I would say don't go with spindle. Go with All Flash unless it's archive.

Most important criteria when selecting a vendor:

  • Supportability
  • Performance
  • Scalability.
Disclosure: I am a real user, and this review is based on my own experience and opinions.

Add a Comment

Why do you like it?

Sign Up with Email