We use it for our EHR. We have 4,000 users who need to have access to a very large EHR called Epic. We are sharing a cache database through AIX servers.
All-Flash Storage Arrays Cache Reviews
Showing reviews of the top ranking products in All-Flash Storage Arrays, containing the term Cache
NetApp AFF (All Flash FAS): Cache
We are looking forward to the all-flash NVMe which is coming out.
Going forward, I would like improvement in the response latencies, capacity size, cache, and controller size. It also needs more fine tuning in regards to all-flash and AML workloads.
The primary use case of this solution is for its speed. We're using the AFF as a cache disk. We have terabytes of data that we have to move quickly off a system. The only way we could do that is with the 40 gig backbone that all-flash array provides and the speed of the disks.
Two things have happened with stability. Number one, the platform that renders the file system is so much better. It's ONTAP and NFS, they're much more superior. The stability of the file system is much better. Behind the scenes, the cache is better, the CPUs are better and of course, there are no spin discs, so it's all flash. That is way more stable than what it used to be. Coupled together, the stability is maybe six to seven hundred times better now than it used to be ten years ago. That's just the way it works now.
This product was brought in when I started with the company, so that's hard for me to answer how it has improved my organization. I would say that it's improved the performance of our virtual machines because we weren't using Flash before this. We were only using Flash Cache. Stepping from Flash Cache with SAS drives up to an all-flash system really had a notable difference.
Thin provisioning enables us to add new applications without having to purchase additional storage. Virtually anything that we need to get started with is going to be smaller at the beginning than what the sales guys that sell our services tell us. We're about to bring in five terabytes of data. Due to the nature of our business operations that could happen over a series of months or even a year. We get that data from our clients. Thin provisioning allows us to use only the storage we need when we need it.
The solution allows the movement of large amounts of data from one data center to another, without interrupting the business. We're only doing that right now for disaster recovery purposes. With that said, it would be much more difficult to move our data at a file-level than at the block level with SnapMirror. We needed a dedicated connection to the DR location regardless, but it's probably saved our IT operations some bandwidth there.
I'm inclined to say the solution reduced our data center costs, but I don't have good modeling on that. The solution was brought in right when I started, so in regards to any cost modeling, I wasn't part of that conversation.
The solution freed us from worrying about storage as a limiting factor. In our line of business, we deal with some highly duplicative data. It has to do with what our customers send us to store and process through on their behalf. Redundant storage due to business workflows doesn't penalize us on the storage side when we get to block-level deduplication and compression. It can make a really big difference there. In some cases, some of the data we host for clients gets the same type of compression you would see in a VDI type environment. It's been really advantageous to us there.
The most important features are the IOPS and the ease of the ONTAP manageability.
The deduplicate process is performed in the cache before it goes to storage, which means that we don't use as much storage.
The versatility of NetApp is what makes it really nice.
Pure Storage FlashArray: Cache
We are doing a project in tandem with Boeing to develop a security solution for their Oracle databases. We've been doing it in the VMware virtual solutions lab, which is back-ended by Pure Storage. It's a very complex project. Pure made it fast enough that we could cycle through the things that we needed to cycle through to get it exactly right. We were able to do so a lot of times, to rev it enough to get it refined to where the process was exactly right every time. There's no way we would have had time to rev it that much had it been on anything slower.
It helps simplify storage. When you're running Pure all-flash, you don't have to do a lot of the old Oracle best practices. You don't have to worry about putting log files on a different disk channel than the data files, and those types of issues. As long as you don't max out the bandwidth of your connectivity, your Fibre Channel, then it doesn't matter. That has pushed the bottleneck down to the connectivity to the storage, as opposed to the different spindle groups on your storage. That has made it vastly easier to do large volumes, rapid provisioning in databases, without taking a performance hit.
We like the data reduction rates. That has been really helpful. You get 4U of Pure Storage replacing something like two racks of spinning disks. One of the things that has contributed to that are the data reduction rates. Not only that, it helps dramatically speed the read coming back in, because you don't have to read it 400 times. Actually, the write doesn't hurt anything either because the write goes in once and then it gets deduplicated and that's that. It does help speed I/O because then everything is coming right off the front end of cache. Certainly, in terms of space, it's probably the most helpful.
HPE Nimble Storage: Cache
We want rock solid stability, making sure all of our customers have complete 100 percent uptime, which is our goal. Nimble achieves that pretty well.
InfoSight is a regular part of our weekly routine. We don't use it every single day, but we do check in on things every once in a while. Luckily, with Nimble, you can forget it and you don't really have to worry about it. However, if we do need to look into an issue, we definitely use InfoSight.
InfoSight has increased the availability of analytics and our ability to quickly get to them. We can migrate things faster. We can pull stuff out of production. We can restore from backups more quickly with the Nimble system than anything else.
We have had a couple of networking issues, temperature alarms, and a few things in our data center going on where InfoSight will scale back our utilization (or whatever) in order to keep us productive and up.
InfoSight has enabled us to get our servers back up faster, especially on the back-end. We have instant recovery. We are able to access that fast storage within seconds, which is very helpful. It enabled us to get service back up in a minute and a half.
The solution has improved our throughput tremendously. It has been on-demand access with 10 gig fiber. The disks, even though some of them are spinning discs, are rated, or in Nimble's little custom config in the way that it hits the cache first, then throws it off to the cold later. This is perfect, and it is great. It has improved everything.
HPE 3PAR StoreServ: Cache
- Four-node performance
- No split IO groups as on IBM SVC clusters.
- Easy tiering (with a small % of cache) did a good job in a large scale environment of 1000 VMs on 350TB external Monitoring, giving a detailed dashboard. A nicely virtual appliance for remote callout support to HPE services.
IBM FlashSystem: Cache
I would like to see an improvement in the handling of large amounts of rights. An automatic flash system that doesn't do compression or deduplication will flush through the rights directly from the host to the flash modules. It doesn't keep them in the cache. For compression and deduplication systems, they have to do compression, deduplication and the memory and cache for the controller. So they have to keep the data there otherwise you will find yourself stuck with performance issues.
Huawei OceanStor: Cache
On a scale from one to ten, I'd rate it at a seven.
I gave it a score of seven based on my experience with other storage solutions. We've used NetApp, HP, and EMC. We've used quite a number. In fact, I gave this score because of the support. It is quite a high score. But frankly, we're still going through the steps of implementing. We have our main database there. Now we are going to optimize further. It's not only because All-Flash costs money. We should use it optimally. There is a tendency for users to put all their data on the All-Flash. We also have cache solutions whereby a part of the data is on the All-Flash. We are trying to convince users that they don't really need All-Flash for all their data, for all their applications. Maybe in more time, I can compare and give you an idea, but not yet at this stage. I can't tell you which features we need, not yet at this stage.
Lenovo ThinkSystem DE Series: Cache
You can not buy a lot of options for these devices. There are a lot of things that it does not do. Some things that it does not do that we would like it to do include easy tiering. If you have got spindles and you want to cache a couple of terabytes of storage on SSD, that would be something we would like to see that, currently, it does not have the capacity to do.
The thing it comes down to is that Lenovo needs to add some more of the software features that would allow the ThinkSystem line to compete with other products that we sell. Other than that, it is what it is.
HPE Primera: Cache
I work with an HPE authorized partner in Malta and we offer storage solutions for customers. HPE Primera is one such product that I have experience with.
We have noticed that these days, most of the customers are implementing a solution that is a hybrid between Nimble and Primera All-Flash. There are both spinning disks and flash, where flash is used as the cache, which makes the price more competitive.
The customers are primarily using it for disaster recovery. They have their cluster and they are replicating one another to provide business continuity and disaster recovery applications.