IBM Spectrum Virtualize Reviews

It enables us to virtualize the entire data center for the Internal Revenue Service. We have about seven or eight terabytes of data behind SVC right now. Before we took this over as a managed service, they had about 120 silo storage areas. Right now, everything is virtualized behind SVC.

It is easier for management with end of life erase. It now takes us less than two-three days to decommission the erase. It has given us an increase in the usage rate on the SVC. It has gone up from probably 30-40% on the silo erase to about 85%.

We have reduced a lot of power and cooling requirements in the data center. We have optimized as much as possible. Right now, it's running very smoothly.

I would like to see data obligation and some features like cluster grouping. I have seen that on the roadmap for the last two years, but it hasn't come out yet. They are still showing it on the roadmap, so I think that it's going to come out soon.

We've had some issues with the production rate behind SVC. We had problems with back-end storage areas having double disk failures. It took a while to recognize errors, due to some of the environmental settings within the data center. When the temperature increased, we started seeing more drive failures on the back-end. But over the last 18 months, we've had zero downtime across the board.

Scalability is very good.

Technical support is very good. We touch base with the IBM house development team in the UK once every quarter to review the roadmap. So it's very good.

The initial setup was straightforward.

We evaluated the Hitachi RV series. I think it was Z100 at that time, and IBM SVC. We tested three solutions and chose IBM SVC to virtualize the environment.

We were looking for stability, such as in the features search and the ease of manageability compared to other products. These products included EMC, Hitachi, and Red Hat.

It's easy to implement. Just become a partner with IBM and you will be successful.

To be able to seamlessly migrate storage sub-systems underneath Spectrum Virtualize.

Being able to migrate storage without impacting the client.

The ability to migrate between clusters, like between Spectrum Virtualize clusters, seamlessly.

Very good. Very stable.

Scalability needs some work in some areas. The number of volumes needs to be increased drastically.

Technical support is very good.

We were using Direct Attached DS8000 before and we switched for scalability.

It was pretty straightforward.

Don't go with EMC. Go with IBM. Go with this product. Absolutely. It's a very good product compared to the competitors.

It's flexibility. It gives us the ability to move the data across multiple storage devices. That makes our lives as the implementers easy during data migration.

It makes data migration straightforward. And on the customer side, there is downtime, but it is very minimal.

I would like to see some features from the Spectrum Control, such as long term reporting. Right now, the reporting on the Storwize and Spectrum Virtualize is only for a few days. You really need another plug-in, like in Control, to get the historical performance data.

The only thing I really don't like about Spectrum Virtualize is the application area. But it has nothing to do with the product. Most application issues come from the link provided by the telecoms.

If they could make the application much easier, that would be great. But it's okay.

The stability is fine. I've been working with Spectrum Virtualize for two years. I saw them encounter problems and those problems are minimal. It's pretty much stable.

It is scalable, especially the VH8 and the SV1. You can locally attach storage to the SVC nodes. It's now more like Storwize. The scalability is head-on.

I have been working with technical support. Before moving to a business partner, I worked with IBM. There were good times and there were bad times. But pretty much, they delivered.

I have had this storage platform for 10 years. I saw the progress of IBM stepping away from conventional hardware storage and going to software. IBM did pretty good research and development on it.

When selecting hardware, I look for technical experience.

I was involved in the initial setup. Most of the time, I dealt with the planning. I start with the planning rack and stack configuration, teaching the customers and then do documentation. The implementation was easy.

I did not look at any alternatives.

I would recommend it. IBM is not the best in the disk storage industry, but they pull out some good tricks.

The most valuable feature is the abstraction layer. Basically, it gives us a single pane of glass between our host and our back-end storage, regardless of what the back-end storage actually is.

Beyond that, I would also add the flexibility of the management itself.

It has significantly lowered management cost, overhead, and everything else. We now have better performance as well.

There is third site replication. Right now, we're limited in our ability to migrate data between clusters. Like I said, we had to scale wide rather than tall and continue to protect our data while we migrate. Additionally, if we wanted to set up a third site for additional DR, we don't really have a good option for that.

Overall, I would say that stability is very good.

There are some caveats with scalability, such as volume count. We are highly limited by volume count at this point in time. We have to grow wide rather than grow tall.

There are some caveats here, too. Sometimes it's been very, very good. Sometimes we have had sessions where we're beating our heads against the wall. It depends on the call. Part of it is has do with us getting better with the technology.

It would be really nice to be able to escalate faster. By the time I'm calling IBM, it's already become such an issue that their Level-1 and usually their Level-2 people can't help me.

We did not use any previous solutions that I am aware of, but this one has been lacking since we first brought it in. It's a wide matrix of considerations. Performance, price, support, availability, and scalability. It's a wide matrix.

The initial setup was a bit of a paradigm shift. Once we got past that, it was very straightforward.

We looked all over the place. But we found what we wanted with this solution.

Absolutely. Period. Do it.

• Capability of putting lots of heterogeneous storage systems all in one management pane
• Being able to substitute them out and move between them
• The ability to put it all in one management and fix them into something that's much more usable.

It's just been handy with a lot of customers that have, over the years, developed a mess of storage.

It's all about manageability. The product lets you manage:

• Different sorts of systems
• The same host multi-pathing driver
• The same feature set: snapshots and replication, regardless of what system it is

Spectrum Virtualize gives you the same functionality and all the same management through the same GUI, regardless of whether you are using EMC storage, Hitachi storage, or IBM storage.

I’d like to see deduplication. IBM announced during Tech U that Storwize is adding it in the upcoming Q3 2017 code update-- same as what exists on the A9000. Duplicate blocks of data write references to pre-existing block of data, instead of re-writing the same data multiple times, thereby saving storage capacity.

For the vast majority of the time, it's absolutely awesome. I had a couple isolated problems with reliability, mostly related to replication, but I have been able to work through them.

The scalability is actually one of the great things about the product. It is designed to be able to give you scalability with the virtualization of multiple storage systems. It's very easy to scale.

I've had to use tech support quite a lot of times, because I support a lot of customers. Consultants support a lot of customers. Once you get past Level-1 support, it's generally top notch.

Good luck finding anything that does half of what it does. When looking for a vendor, I look for reliability. I've had lots of good experiences with it. It's been a very reliable product and the feature set is something you can't get anywhere else.

It has been integrated into our production systems, thereby not having as much online storage. We're using the archiving functionality. When I say reliability, I mean that we are operating 24/7. The product enables us to retrieve our client data at any given point in time.

The most valuable feature is its reliability.

They are actually working on one bug we found, which was with flash restore. This was the user interface design for virtual environments. Supposedly, it was going to be corrected in 8.1, restoring it back to the original UID, but it wasn't done.

The stability is excellent.

At this point in time, scalability is excellent. It's very scalable.

I work in Miami, Florida. Our data center is elsewhere. We may have some connectivity issues every now and then, but overall, it's been excellent. I always make certain to have the right phone numbers for support.

I wasn't involved in the initial setup, but I'm involved with the new progression from old TSN to Spectrum 8.1.

We look for reliability of the product itself. It's an excellent product. I would not be penny wise and pound foolish, though.

I would honestly utilize as much IBM as I could. I'm an old system manager in multiple, prior jobs. I would always stick with IBM across the board, especially if you look at the high-end tape units. They will get around to the correct drivers and everything. It's much easier to use all IBM.

If you use someone else's server, whether it is an X86 or whatever, you get finger pointed if something doesn't exactly work right. This is especially the case with tape drives, and especially if you were using Jaguar, which was the old type. They came off mainframe. It was very high end, very costly. I would stick with using IBM servers, even IBM storage like XIV if you want to go with something less costly than a DS88.

Make it correct. Make it easy on yourself. Use HPE storage, disc storage, or Dell storage and you will get finger pointing. It always happens. No one's wrong. Even if you get IBM, sometimes they are also wrong.

We had all IBM and we did have an issue when we upgraded our tape libraries, i.e., we had the wrong firmware. That was with IBM. Imagine if we had someone else. It would be a long, drawn-out process.

You may have one or two issues at the same time, and one can mask another issue. Don't go cheap. Have a test system. Never, ever, put something straight into production. I don't care how many things they swear on, or whatever.

You never know because everyone's environment is different. That's the other thing. I don't care if it was AIX, and it's just moving into Linux, you still need to test it. Put it up for a few weeks, if not for a few months. Don't ever go cheap on a test system. If you can have it separate, have it separate from your other actual production servers. In some places, we actually had it in a different machine.

Have a different machine. Never combine. Keep it simple.

Being able to move things around from one array to the other has been the biggest feature for us.

There's a little bit of performance benefit. The flexibility that we get from being able to have different vendors' storage arrays presented as one homogenous unit to our hosts has been the main benefit.

Currently, the newer features exceed our needs. What we really need is resiliency and stability in the features that are out there and do run.

We have been using the system for twelve years.

It's been pretty stable, but we've had some issues. I wouldn't say it's poor, but the resiliency has been less than what we expect from an enterprise class product.

It appears to us, over several decades, that quality control waxes and wanes. That's scary for us, because as a new system comes out and we install it, we really expect it to be an enterprise class product and not to have problems. We expect it all to be bullet-proof, and it hasn't been in our experience.

Scalability is limited, but it meets our needs.

Generally, technical support is good.

It's been a long time now, about twelve years. I forgot about previous solutions.

The initial setup was straightforward.

I'd tell them that it's a good product and they should implement it. They probably should not run the latest code and they should look very carefully at advanced features.

When selecting a vendor, honesty is one of the first things to consider, and then stability and resiliency of the products. Performance is nice, but the applications that I support need bullet-proof stuff behind it with no down time.

• Reducing our costs: We used to have all tier DS8 storage. Even though we had Spectrum Virtualize in front of almost all of that, we still had almost all DS8.
• We've been able to bring in multiple tiers, flash systems, and V7000s

• We can migrate that data, watch it with the tool, and know that the data is not on the right tier.

• We can migrate that data again, place it in the right tier, and reduce our overall cost.

The most valuable feature is probably data migration so we can bring in back-end systems and swap data out. Our end-users, and even our other system administrators, don't have any idea that we've moved storage around.

It's got full features, so we can compress volumes. We can do thin volumes and we can change them on the fly.

Nobody knows that we've migrated that data around. We can ship it off to our DR site. This is all under the hood of Spectrum Virtualize.

We don't have to worry about what type of block is underneath it at the time. It's all being done at that layer.

A feature that is already there, if I remember correctly, is encryption. I think it is coming out, or it is already there.

That is a key management piece. Right now, we're doing an encryption on the back-end flash systems on the V7000s. It's simple, with just a USB key into the controllers.

The integration would be an option that we would like, but I understand that's not how it's going to be implemented.

NPIV is also coming. I'm not exactly sure what benefit it will bring. Initially, that sounded like that was going to be kind of cool. Even though we can migrate data without our end users really knowing it, they do see a path failure, and NPIV would take care of that for us.

The feature that's kind of missing is getting us up to the point where we can help the application owners see where their data is at, understand it, and potentially help us breakout.

We've used easy tiered functions in the pools, so we're trying to help step that storage down. If they can get visibility somehow into that data, help us further break that down, or better tier and separate out their data, that would be helpful.

I know that VMWare has that function, where they are taking multiple tiers themselves and placing subsets of data, as opposed to whole blocks.

For the most part, stability has been really good. Like anything else, the more you use it, the more times you're going to run into a bug. We've certainly done that.

It seems like scalability is OK. I mean, it's kind of hard for us. We're on a four year refresh cycle. So we stay pretty current with the hardware itself. We are adding in things like compression accelerators, and now with the new SV1 nodes, more cache, and we are going to more adapters. Of course, we're always willing to split out workloads onto separate stacks. So it scales pretty well.

Support has been pretty responsive. Software is coming out all the time with PTFs to fix, so staying on top of that is important.

We've run into the seven three code. Specifically, we hit the cache performance bug. We were right on top of that and had to do another upgrade to clear that.

We've also had a bug with the fiber channel cards. Again, by the time we were implementing, it was a known issue. It was not anything that we've really had to wait on, but not something that we were aware of at the time of implementation.

Get a demo of it. If you haven't seen the product and you have not had somebody step you through all of the features and all the things that you can do with it, then I think it would be really tough to see where adding another set of controllers in front of your storage is benefiting you.

You might be thinking, that's just another hop and it's another delay in getting to my data. I think you will see the value of this solution once you:

• See it plugged in
• Understand what's going to come with being able to move, compress, and virtualize your data in one interface

• Are able to manage all the data there, and not worry about the back end-stuff
• Are able to carve up volumes very quickly to the end users

Integration with Spectrum control and a kind of self-service provisioning is good. It is something we're looking at turning over and then deciding about all the data migration that can happen in the back-end. We can look at that request, and then decide. Perhaps we didn't have enough information when we started, and then we can move it on the back-end. We don't have to worry about getting into the weeds, necessarily, from day one.