IBM FlashSystem Reviews

Storage virtualisation: Use o storage pool to spread I-O load across as many disks as possible, reduces “hot spots”.

Ability to manage third-party arrays and virtualise them: One screen to control multiple arrays. Simplified administration.

Auto-tiering with Easy-Tier: High usage volumes and parts of volumes are automatically moved to high performance disks in manageable “chunks”. This means volumes with heavy I-O usage are are moved to SSD for the best response. Zero admin overhead.

Using V7000, the storage administrator has complete visibility into the I-O load and response times of the volumes, storage pool. Together with “Easy Tier”, this means volumes with high workloads are automatically moved to the highest performing disks (SSD) in the array. There is almost zero administration required once the storage pool is established.

Data migration: The migration wizard made migration from old to new an absolute doddle. Something that historically had been “hard and difficult” became a background task, once the old array was managed by the V7000.

The replication of data between IBM Storwize arrays in V1 of the product had some major limitations. A lot of the limitations were overcome in V2 of the product. Some improvement can still be made in scalability using clustered V7000 nodes.

One area for improvement is in the GUI, where host clusters are not properly dealt with. With Hypervisor host clusters, all hosts must see the same volumes in the same order. Using the concept of a “host-group” has been around (even with IBM) for many years, so why not with the V7000?

Four years.

Some isues were encountered during upgrades in the early firmware versions. This has now been resolved.

The V7000 can scale up to 10 enclosures (disk shelves), each up to 24 disks per enclosure. A cluster of 2 x V7000 has twice the capacity.

Using the V7000 you must be careful not to overload the CPU and disk backplane. When planning to implement any storage array, you must plan your workload carefully. IBM have a superb tool, Disk Magic, that predicts the I-O performance of the array configuration you have selected. This is an invaluable tool in planning any IBM storage array.

I was lucky to have a very good relationship with the local IBM storage support experts. The implementations were done under IBM guidance and, knowing this, any issues that came up at my sites were quickly dealt with. The support staff knew the arrays had been installed with expert guidance and had no inherent issues in design.

Previous storage solutions included the HPE 3PAR array family. We switched to IBM for 3 reasons:

• Upgrade cost
• Complexity with administration tasks and day-to-day running
• Support costs

The IBM V7000 initial setup was straightforward, as I had engaged IBM support early on and was given very good advice on how to size, configure and implement the array(s), including data migration from the old to the new.

It pays to go back and get the best price you can from your supplier. The first offer is not always at the best discount.

Ensure that all the features you need (including replication) are included in the license options.

Several vendors were evaluated in addition to IBM: HPE, NetApp, EMC.

1. Download the IBM Redbook on best practices to implement.
2. Engage IBM support or your integration partner/reseller at a detailed level.
3. Know what you want to do and document your requirements.
4. Investigate upgrade costs and five-year maintenance options.

In the Unified model it supports SAN and NAS solutions very smoothly. Also, there is a back end command line interface which is very comfortable for scripting, especially for snapshot and business copy solutions.

V7K has better I/O performance than some other storage, such as HPE XP12K, HPE EVA (4K and 8K) and IBM DS (3K and 4K). Because of this, I have seen improvements in long-term business operations (like billing run in a telecom company and end-of-month operations in a financial institution).

It should increase sociability with respect to number of disk shelves and capacity.

Overall, five years.

No.

No.

Eight out of 10.

Switched to decommission older hardware (XP12K and EVA from HPE).

Very simple setup, especially for FC SAN network.

I cannot comment on this as I used it as an IBMer.

I compared the product with IBM DS8K. Due to a huge gap in price, and not having very high-level needs, I selected V7000.

This is highly reliable storage. Other than not having very high IOPS requirements, RAID5 would be good enough for application and database LUN. But before creating the RAID group/array, the implementer should have proper projections.

• Thin Provisioning
• Flash Copy
• User friendly management interface

We run our business applications on it. Also OS was booted from SAN.

Product support is restricted to IBM only. It must be decentralized to IBM partners as well.

Six years.

Yes, we faced power supply failure, Multiple SAS drives failure, and at the end both controllers were dead.

We are able to add additional drives easily.

One out of 10.

No, it was our first experience with any kind of SAN.

Initial SAN configuration was smooth.

If you are planning to purchase a new SAN then go for all flash. Do not invest in a hybrid SAN.

Yes, HPE and Fujitsu.

Please make sure that you will be getting proper after-sales support from IBM.

• User friendly interface
• Robust hardware
• Good support

VMware and Hyper-V datastore (with Fiber Connect). All virtual systems are running on the Storwize V7000 storage in our company.

Deduplication and DRC synchronization. Deduplication is small, limited. We need better deduplication and disaster recovery.

Since 2014.

Good stability.

No.

Global support is good.

Yes, I myself have used. The customer decided to switch.

Yes, it is easy in terms of management and installation.

Yes, I evaluated HPE, Dell EMC.

Check IOP values and disk speed. Check often for updates and firmware. Make absolute SNMP settings for warnings and alerts.

This was a great step forward from the XIV in next generation hardware for the customer. We're seeing tremendous data reduction for many workloads. It's not a 100% move from XIV for all workloads, however, but the customer is very pleased, for the most part, in their environment.

Benefits for us include lower overall cost, higher density, lower data center costs, and reduced total TCO, as well as moving to an all-flash solution versus a flash/spinning disk solution.

The HSM management console GUI still need some enhancements. It is still a bit awkward in navigation, and some functions/features are not yet fully integrated into the GUI support, but only via CLI. We also need to get reporting down to individual pool and volume levels, rather than just overall system levels. Customers need a more granular view of their individual business usage as well.

We've had some issues with stability, such as a couple of hardware failures and a couple of code bugs. They were addressed, but they caused initial customer concern.

Scalability is quite nice, with being able to support over 1.5TB effective in one rack is a nice selling feature. One of the early issues was that they weren't at a code release where we were able to add new elements right away in a non-disruptive fashion. We had to wait for a couple of code releases for that. We've gone beyond that now. At this point, scalability is no longer an issue.

Technical support has been great. They have been engaged from start to finish.

Regarding previous solutions, this was more of a technology refresh decision. A customer was heavily invested in an older technology that they were very satisfied with, but they wanted to continue to make the move to flash technology. This was the next logical era of evolution.

The initial setup was a little more difficult than the predecessor XIV, but it's a new product, so a learning curve is to be expected. As it is becoming more common in the field, it should be getting better documented and more straightforward to deploy. We are seeing new and continual updates to Red books and best practices documents now.

We didn't really consider alternative solutions, mainly because there aren’t any out there that can touch the A9000 family. From our standpoint, we're an IBM business partner. We value reliability and performance above everything else. Price is a close second, but reliability is key.

Read anything and everything you can, such as documentation manuals and best practices that are coming out. Stay plugged into the IBM communities and leverage services if this is your first engagement for capacity and sizing.

Auto Tier is 1M-iops.

It improves our number of TPS.

Flash Manager needs to be included.

2 years.

Only battery for controller.

None so far.

If 10 is the highest, I would rate them at 9.

We were using a V7000 before, so only upgraded.

Almost straightforward and the same as doing a V7000 series.

Licensing has already everything you need like replication, but you need to consider the license cost for the SAN switch.

EMC Xtreme IO, HPE 3PAR, Nimble.

Consider which type of disk to use: All flash, High Capacity Drives, SSDs.

I/O speeds. This products in basically a RAID comprised of SSD drives with PCI-Express interface, that you can connect to 16GB HBAs, making this storage one of the fastest available in the market today.

This storage has incredible performance, the I/O is awesome. It is difficult to understand how fast it is if you don’t have one, that’s why IBM gives you a month trial, to install it in your infrastructure and give it a try.
In my organization's case, we use the FlashSystem basically for virtualization (VMware). It's amazing how fast a virtual machine boots. We use it for databases too, and the disk I/O performance is incredible.

This storage is factory configured so it's very difficult and expensive to add more flash modules (disk capacity).

Scalability is the main area which needs to improve.

Six months.

No, no at all.

IBM support is excellent as always.

No, this is my first experience with a 100% SSD solution.

No, in fact it was easier than I expected.

If you are going to buy one, have in mind the amount of capacity you need, and your estimated growth for the following years. As I mentioned above, it is expensive to add more SSD modules.

Not really, it was recommended to me by a friend, and after the first month, my company couldn't live without it.

I recommend adding a pair of SAN switches to the Flash solution, otherwise it's a waste to use as and direct attach storage.

The most valuable feature is the continuity of the user interface. Many of our customers leverage Storwize and XIV storage subsystems. That continuity in look and feel has likely helped the FlashSystem because transitioning to using this new product doesn't require any additional training or familiarization time. Instead, customers leverage their experiences to perform 90-95% of the tasks necessary to get a new FlashSystem 900 running.

It's especially important to be able to walk into a customer site and say, "Hey, you have access to the same user experience that you have with your Storwize system, and/or a very similar user experience to your XIV."

Having a consistent user experience and user interface has helped save our customers' time: we don't have to retrain them for a new user interface, and they don't have the learning curve associated with understanding the nuances of a new interface.

The FlashSystem has really helped improve our customers' I/O intensive workloads. Many of our customers leverage Flash for database workloads and appreciate the extra performance boost it gives.

A lot of the beauty of the FlashSystem 900 is that it's very, very focused on what it does, and it does it well. I don't really think there are additional features required.

The FlashSystem 900 is a relatively stable platform. While any codebase will have some issues and require patching (nothing is perfect), the codebase on which the FlashSystem 900 is built is solid.

As a building block, the FlashSystem 900 isn't really intended to scale out to a large degree. Instead, it's intended to be used as a block with which to build out a scalable solution (such as with an A9000/A9000R or a V9000).

IBM's support for storage hardware is usually excellent.

IBM's storage technical support is usually excellent.

I have been involved in the initial setup with a number of customers. It's very straightforward. It's a lot easier than the old days.

If you are looking for a great flash solution that can be virtualized or utilized as a building block for a scalable flash solution, or if you're looking for flash for a very specific workload, then you may want to consider a FlashSystem 900.

It's probably the most straightforward piece of hardware you can get. The user experience from the interfaces and the price points for the performance a really solid.

I'm an IBM Business Partner, so there's really only one vendor that we go to for our storage. However, even if I weren't an IBM Business Partner, having a pretty stable track record with their products, ease of use, and value delivered for price -- I would probably still recommend a FlashSystem 900 for the use cases I stated earlier.