What is our primary use case?
We are the centralized IT department for a state government and we service every agency in the state. That includes anything from the state police down to DNR, parks, unemployment, and DHHS. There is a wide variety of use cases, but the big hitters on it are Oracle and SQL databases.
It's on-prem. It's in two different data centers that are 60 miles apart and we're doing a synchronous replication between the data centers.
How has it helped my organization?
There are so many ways it has helped. It provides efficiencies through compression and it provides high availability through its solid-state drives. We literally turn it on and it does its thing.
When it comes to storage provisioning, a lot of it has been automated. This was true even prior to PowerMax, back with the VMAX. The days of provisioning the mapping and masking, and doing all those things manually, are over. A lot of that is automated through their tools. Overall, that automation is saving us about four hours a week.
What is most valuable?
What is most valuable to us is the fact that it has multiple engines, and each of those engines works in conjunction in a grid environment. That's important to us because we have so many different use cases. One example might be that a state trooper pulls someone over at 2 o'clock on Sunday morning and wants to go into the LEIN system, which is the law enforcement information network. He wants to see who this person is that he has pulled over and gather as much information as he can on that person. We can't predict when he's going to pull someone over, nor can we predict when backups are actually going to be taken against the volume that he's going to for that information. The PowerMax allows us to do backups of that volume at the same time that he is looking up the data he needs, and there's no impact on performance at all.
The performance is very good. Our predominant workloads are all less than 5 milliseconds and it's most common to have a sub-1-millisecond response time for our applications. In terms of efficiency, we've turned on compression and we're able to get as high as two-to-one compression on our workloads, on average. Some workloads can't compress and some can compress better, but on average, we're a little bit more than two-to-one.
The solution’s built-in QoS capabilities for providing workload congestion protection work pretty well because we actually don't even turn on the service level options. We leave it to the default settings and allow it to decide the performance. We don't enforce the Platinum, Gold, or Silver QoS levels. We just let the array handle it all, and it does so.
We also use VPLEX Metro, which is a separate service offering from Dell EMC. It does SRDF-like things, but it's really SRDF on steroids. Of course it copies data from one data center to the other, but with the VPLEX, not only does it copy it synchronously, but it also has coherent caching between both data centers. That means we are literally in an Active-Active mode. For instance, we can dynamically move a VMware host that is in one data center to another data center, and we're not just doing vMotion with the host. The data is already in there at the other data center as well. It's all seamless. We don't have to stop SRDF and remount it on another drive. It's already there.
For how long have I used the solution?
We have been using Dell EMC PowerMax NVMe ever since it was brought to market, so it's been about three or four years.
What do I think about the stability of the solution?
It's rock-solid with 100 hundred percent uptime. We've never had a disruption on our PowerMax platform. It's high availability. And we can make changes, such as upgrading the code, while it's running. There's no such thing as going offline to do a service or maintenance procedure. It's all done online and the customers are working away at the same time.
What do I think about the scalability of the solution?
The scalability is great. VPLEX is something like a federation for all our PowerMaxs. We will put a PowerMax in, give it all to VPLEX to manage, and we're good to go.
We typically see a 10 to 20 percent growth rate, year to year. To keep up with that, in a multi-petabyte environment, 10 percent is quite a lot. We buy two a year, and that's a conservative estimate.
The fact that PowerMax provides NVMe scale-out capabilities is important from the standpoint of its internal workings, but the customer data doesn't really go on the NVMe technology. At this point, we don't have any use cases for NVMe performance for any of our applications. But that will change in the future. Everything is going to go to in-memory. Compute and storage: everything's going to be on a chip.
How are customer service and technical support?
Their technical support is really good. We are using one of their monitoring tools and it phones home to the "mothership" in Massachusetts. That means they get real-time alerts or performance indicators. If a drive has exceeded a threshold five times in the last week, they will actually come out and preemptively replace that drive before it fails.
Which solution did I use previously and why did I switch?
We were a VMAX customer, so when they changed their service offering from VMAX to PowerMax, that's when we started adopting it. In a sense, PowerMax is the first of its kind for us. But we have been a long-time customer. We started with their DMX almost 20 years ago.
How was the initial setup?
For us, it's straightforward to set up. We've been doing this for a long time, so it's really easy for us to set up a new array in a data center. We had one that hit the dock about two weeks ago and it's already up and running and provisioning to customers.
NetApp will say, "Well, that's two weeks. We can come in and do it in one day." But we explain, "No, you can't because there are internal processes that we have to go through." Every piece of equipment we get, even the PowerMax, goes through its paces. We don't just turn it on and hope for the best. We check and double-check all our configuration settings. But overall, PowerMax is easy to set up. They configure it at the factory, deliver it, put it in the data center, and then we hook it to our Fibre Channel fabric and Ethernet fabrics and we're good to go. Competitors will say, "Well, it's so much easier to migrate from one array to another on our platform, versus the Dell EMCs." That's not necessarily true. We have to look at what they are actually measuring and whether we are comparing apples to apples.
With VPLEX, we can do migrations on-the-fly, live. It's no longer a six-month to one-year effort to get off of one array and move to another. We just bring the other array in, present it to VPLEX, and VPLEX takes it from there.
For a new deployment of one PowerMax, we need one FTE. On a day-to-day basis, to manage all of our PowerMaxs, we need three FTEs. But that is across two different data centers with a total of 10 PowerMax/VMAX units. It's a pretty big installation. Across our organization we have 55,000 employees. Since our HR is on this solution, and that's how people get paid, it's like we have 55,000 people using it, in a sense. Most access is through an application, but in another sense, it's used by pretty much everybody in the state.
What was our ROI?
On a typical purchase, the ROI is four years. That's when we get our money back. We charge for our service and we have a rate per GB. Our business model is set up to only recover our costs because we're government. We can't make a profit on it.
What's my experience with pricing, setup cost, and licensing?
One area for improvement, one that everybody always comes to, is price. Although we get a good discount through Dell EMC, it's still quite expensive to purchase these big arrays. I buy in volumes of petabytes at a time. It's not unusual for me to have a $6 million spend. While that is petabytes of data, it always raises eyebrows when you spend that kind of money. But what I ask those raised eyebrows is, "Okay, fine. Which of the agencies in the state do you not want to give more storage to? Everybody's using it."
Many competitive vendors will come to us and say, "We have a study where we went into a company and we were able to reduce their costs by 600 percent." Of course, these are salespeople and they're speaking to two levels above me, and they buy into that and say, "Yeah, let's have them come in and talk to us." They come in and talk to us and when we get to the stage where we say, "Here's a typical configuration. Give us a quote for that type of configuration." When we compare it to the cost that we're getting from Dell EMC after the discount, it's plus or minus 5 percent. There really isn't that big of a delta compared to our pricing. This is a high-end device. For us, the pricing doesn't make Dell EMC uncompetitive.
Which other solutions did I evaluate?
NetApp and Pure Storage are the biggest ones we looked at for block storage.
For other storage, like file, print, and object , there are a dozen others that are always trying to differentiate themselves on price. They want to do a proof of concept and we do those with them. But what I'll tell them up front is, "I know your products are great. They're going to work great in our lab. You don't really have to send me a piece of equipment for me to test it. I know it's going to work. You guys wouldn't be in business if they didn't work. So let's get down to the cost of it." And when we get to the cost of it, it's just not compelling enough to make a switch.
But as far as features go, I don't find there is a huge difference.
What other advice do I have?
The biggest lesson I've learned using PowerMax is to trust it. For example, with the QoS, don't try and overthink this. It's engineered to take on diverse and disparate workloads. Put it in, watch it for a little bit, and if you don't absolutely need to turn on all the QoS, don't. Let it do its thing.
Don't be shocked by the price per GB. Look at your cost of transactions or IOPS. The days of looking at storage as so much per GB are over. It's how much workload you can pass through that storage device.
Overall, PowerMax is ideal for storage for enterprise-level, mission-critical IT workloads. That is really its strength, as is its ability to handle disparate workloads. I wouldn't use anything else for these high-end, critical workloads.
Which version of this solution are you currently using?