What is our primary use case?
We're currently running two 3PAR 7200 storage units in high availability. We have three workload tiers. We have Nearline, FAST class, and SSD. Our primary ERP system is an Oracle JD Edwards running on Microsoft SQL Server 2008 R2 that is all on SSD. Then, we have other workloads for our barcode. Our engineering solutions are running on FAST class, and then most of our traditional file and print, storage, and workloads are running on Nearline SATA. Also, have two 4200 LeftHand SANs in the environment. I put very low priority VMs on those two LeftHand SANs. They are minor application servers. They don't need a whole lot of performance. However, the LeftHand SANs are now seven years old. The 3PAR SANs are now five years old, and I have to replace everything in 2020, and I'm looking at HPE SimpliVity, Nimble, and potentially 3PAR as the storage architecture for that environment.
Our JD Edwards, which is our ERP system, that is critical. Also, our barcode scanning, because we do a lot of barcode scanning out in the shipping and manufacturing warehouse. Our accounting system is part of the JD Edwards too. All of that is on the SSD. We're currently evaluating whether we upgrade to JD Edwards 9.2 or if we deploy Microsoft Finance and Operations. If we go with Microsoft Finance and Operations, that'll be totally in the cloud, and I'll be able to carve a third of my storage requirements out because it will no longer be necessary to run an on-premise ERP solution.
My directive when I was hired in 2016 as a direct IT manager versus an outsourced IT manager, as I was when I started in 2014, is anything and everything I can take to the cloud goes to the cloud. If I do that, it reduces the need for all SSD on-premise, and that's actually what I'm trying to get to, because I'd rather utilize Microsoft Cloud, Azure, Office 365, and Dynamics 365. I want to utilize that cloud for my performance, whereas on-premise traditional file, print, and storage doesn't really need SSD.
How has it helped my organization?
In the deployment of virtual servers, I can have a new VM up and running in 15 minutes, run the patches, then done. I routinely fire up base images that I have for my servers: Server 2008 R2, 2012 R2, and 2016. I routinely fire those base images up and do all the updates, then prep them again for cloning. With 3PAR, we definitely have the performance to do that. Those images I do keep on SSD just to have that performance to deploy a new VM.
It has improved our throughput. We went from a 1G backbone on our LeftHand SAN, and also the IBM SAN was not redundant, but we put in a 10G backbone with 10G fiber which truly increased our performance.
What is most valuable?
The high availability is awesome.
The reliability: In five years, I have had one drive that has failed, which is not so much 3PAR as much as it is HPE. I have depended on HPE servers since it was Compaq servers.
What needs improvement?
One of the things I like about the Microsoft operating systems are Microsoft's built-in backups. It's not elegant nor real pretty, but it just works for a single server or single VM. What I would really like to see from HPE is backup built-in, not snapshots nor replaying snapshots, but true block or file level backups integrated into an HPE platform program.
Today, I'm using StorageCraft, where I can have a VM recovered in 15 to 20 minutes. I run a continuous restore point on three of my primary domain controllers. I run a continuous restore point against my primary Microsoft SQL server. So, I always have that continuous, but it's taking up so much storage space that I keep running out and having to add. I need something better, as I've been doing this for a long time. Maybe having Arcserve Backup or Seagate Backup Exec more integrated into the hardware solution.
What do I think about the stability of the solution?
in the last five years, we've only had one drive failure. We have run out of space on our Nearline storage, because our Nearline storage was set up for the Remote Copy groups. I'm actually going to be working with an HPE storage architect next week to redistribute the ARC copy groups. Most of that is low tier that I'm not worried about for high availability, but for our SSD storage and FAST class storage, the high availability is there. I have a six node cluster that we are splitting to two sites on one campus. If I have any failure, everything fails over in a heartbeat, then nobody will hardly notice.
In the past, I've been comfortable running HPE servers for more than 10 years. However, I no longer find that to be cost-effective, especially as a customer, because running something that long obviously induces too much risk. Now, I'm on a three to five-year replacement cycle, and the current environment was installed five years ago. I have to get it replaced next year.
How are customer service and technical support?
3PAR technical support has been outstanding. When I ran into some issues with my Nearline drives last year, they helped out with that. Every time that I have had to update the OS or firmware, I opened up a case with 3PAR support, and I get somebody right away. Then, we schedule time and are able to do live updates with no downtime. This is huge and critical for me.
Which solution did I use previously and why did I switch?
I told the company that they needed to invest in 3PAR. They had one IBM DS 5600, a SAN that was huge, but was only eight terabytes, at the time. They were trying to go with the two LeftHand SANs to replace the IBM, but the LeftHand SANs IOPS could never succeed the IOPS from the IBM SAN. Then, the reseller that was helping them was let go and new management was brought in. That management decided to outsource IT to the company that I worked for. I was a huge HPE partner at the time.
3PAR has increased our performance over our old IBM SAN that was put in around 2009, prior to my time, which was 1G fiber and all SAS drives. The performance of the 3PAR, with its SSDs and controller nodes, was vastly superior to that older IBM. The HPE LeftHand SANs were all SAS and SATA, so the 3PAR SSD performance was just off the charts.
How was the initial setup?
Anything new can be complex. There were some things in the initial deployment that I was not happy about. One of my directives was, "However, it's configured, ensure that it can never be overprovisioned." That one key thing was overlooked.
This is why I had to have a support call last year, because it actually became overprovisioned and I had to move some stuff around. We're still sort of in that with the Nearline storage, and that's why I'm having another architect work with me next week, so we can redo some things. I've had to move a lot of what was on the Nearline storage over to the LeftHand SANs. If I hadn't had those LeftHand SANs just sitting there with 20 terabytes, I would've been in serious trouble. That was my one point with the deployment team, everybody thinks, "If you do this, you do this, and do this, you'll never get overprovisioned," but I've been doing this long enough, and going back to Compaq StorageWorks, if you're not careful on how you have Remote Copy groups, your redundancy setup, thin provision, thick provision, lazy zero, and eager zero, then you will get overprovisioned at some point. You will lock up on a SAN tighter than a drum.
What about the implementation team?
I was actually outsourced as the IT manager. Starting in February of 2014, I was selected to lead the management of all the company's nine locations in North America, Europe, and Canada, working with HPE at the time. After we selected to go definitely with 3PAR, HPE helped do the whole design, and I approved the design. It was actually deployed in August of 2014. Everything was deployed to my specifications.
What was our ROI?
What's my experience with pricing, setup cost, and licensing?
For the entire six node cluster and the two 7200 units plus the Brocade Fibre switches, we financed it through HPE Financial. It was $850,000. We leased that and paid it off in October of 2018.
I have to renew support in October for the existing solution. That will cost me roughly $50,000 this year, which in the grand scheme of things, $50,000 is not that much compared to paying $850,000. However, it will be the last year that I'll be able to get direct support from HPE. Therefore, product has to be replaced in 2020.
Which other solutions did I evaluate?
I worked with the team from HPE. We looked at their solutions, and selected 3PAR. We also looked at Dell EMC and Nutanix. At the time, when evaluating Dell EMC and other SAN products against 3PAR out-of-the-box, 3PAR just was vastly superior to anything else.
Working with the HPE team back in early 2014, they proposed the design. As it turns out, the SAN that we deployed in 2014 was the largest 3PAR SAN in the state of Wisconsin.
I just have a long history of HPE. I've tried Dell EMC and IBM before. IBM sold its server products to Lenovo. I always come back to HPE, especially the ProLiant brand, just because of the reliability and consistency through all the generations. You can look at a ProLiant 1000, never having seen one before, but if you know a Gen 9 or a Gen 10 today, if you could find a Proliant 1000 that was still operating, then you would know how to go in and configure it.
It is this type of consistency that keeps me with HPE. Dell EMC is all over the place. Lenovo has reliability issues.
What other advice do I have?
Look at HPE's roadmap for 3PAR, SimpliVity, and Nimble. Do your research, then pick the right one that works for you with the future that you envision.
I'm highly interested in using InfoSight going forward. One of the things that I have always tried to do is get to where I had just one dashboard for everything from managing from the desktop up to my Internet perimeter security. I want to look more at Aruba Networks and Cloud services to see how that might be able to help me integrate my WatchGuard perimeter security. I'm looking at Commvault and switching to Commvault for my backups, because eventually I just want one dashboard that shows me everything: servers, storage, switches, access points, and security perimeter points. That is the platform that can get me there.
That's one of the reasons why I'm at the conference is to check out HPE GreenLake Flex Capacity.
Biggest lesson learnt: Don't be absent during the design phase.