Please share with the community what you think needs improvement with Dell EMC PowerMax NVMe.
What are its weaknesses? What would you like to see changed in a future version?
The improvements made to the product line over the generations has made PowerMax a gem. Nothing being perfect, the improvements that come to mind would not be specific to the physical product, but instead on the support and management side. Support of the product can be slow and an administrative challenge: planning, scheduling, and overseeing data center access for a Dell EMC rep. One improvement could be to enable a self-maintenance option. The requirements that we go through to get Dell EMC onsite to replace failed drives, power supplies, and other small redundant parts can be unnecessarily complex. If simplified, they could send us the parts, then we could replace them much faster, more easily, and truly within the SLA parameters. We have had performance/availability issues in the past with the management server/application, Unisphere. Upgrades to the platform could also be difficult and even fail. However, the most recent version released last month had been the first in a long time that was successful. Therefore, we are hopeful those past software issues have been addressed.
The visibility within the storage resource tools or understanding the utilization of the SCM memory have been pain points. We know they are being used, but it is hard to actually see them within the actual GUI. Firmware updates are a bit painful because you have to involve their support, as opposed to having the ability to do it yourself. This is probably for the best because you don't want something to go sideways while being the only person working on this and not having external support for it.
It's a relatively new product, but for the next release I would like to see higher bandwidth on the front-end adapters. This would allow even greater scalability for critical workloads and consolidation for non-critical workloads. The hosts may not require that level of I/O performance today. However, it allows us to scale physical non-cloud environments without large investment.
The product would be better served if there was a slight reduction in price at the moment due to the marketplace. People haven't got as much money. If they could offer more of a discount, it would help their customer base out quite a bit. Even if it was just in the short term, it would make a big difference. If the solution had more power-saving capabilities, it would be quite nice. The solution could benefit from even more speed and increased redundancy and flexibility.
The main feature that I personally want to see is the possibility to upgrade to the next generation without changing all the components and just change the engine, relying on the compatibility matrices between two different generations. Meaning that we could just keep the enclosure and upgrade the engine, integrating the enclosure to the existing pool, then adding automation tools for orchestration. When you move from VMAX 200K to PowerMax you swap Array. Or DELL EMC must give to the customer the ability to reuse component to the new Array. For example with IBM Storage like Storwize you can reuse enclosure from Gen2,2+ on Gen 3
I would like to see more cache because we are limited to two terabytes of cache now. More cache would potentially help drive better improvements in performance. Also, when it comes to Oracle and database workloads, data reduction could be a little bit better. Some of the competition, like Pure, have post-processes which do additional deduplication and compression on the backside; everything is in-line and then they do a secondary process. It would be a good option if you could start getting 5:1 or 6:1 data reduction on database workloads. That would be beneficial.
It would be nice if there was a training course offered by Dell EMC. It would help us use the product.
I started using CloudIQ two days ago, and all it's been doing is filling up my phone with alerts that aren't worthwhile. There is something going on there that the array is flagging things as inappropriate that aren't really impactful. I would like to have Snapchat scheduling and the ability to modify that instead of erase a schedule, then recreate it. There are way better ways to do that. Support for SRDF consistency groups within the GUI, instead of making that the command line. Remove the need for physical or hardwired virtual servers to run consistency groups, instead make the expensive array controllers handle that. The management interface needs improvement. It shouldn't be as hard to do some of the functions as it is. Also, it shouldn't need Windows Servers to run a million dollar array.
There is room for improvement in the replication. It's an important requirement for us.
I am looking for ease in usability going forward. PowerMax is super powerful, but because it's been around for so long, there is some complexity in configuration and getting the right SLAs set up that you want. I feel like this could be simplified. I would like to see some improvements from there to avoid having to hunt and peck through an interface to do something that I feel should be relatively simple. I hear from people on my team that they would like improved reporting. While there are some decent tools for doing reporting, they would like to see a lot more built-in functionality. This way when they are logged into the interface everyday doing basic management tasks, they could also see some statistics on what is their storage pool usage and will be their projected usage with their current data growth. They want to be able to see more detailed stats on how they are using the system and have forecasting.
I would like them to continue improving the management tools and continue moving towards a RESTful API versus CLI. They should work with the storage engineers to better tweak the management tools to give them improved visibility into their data.
We are very interested in NVMe over Fibre Channel, which I understand is on the horizon. We would like to see that come to fruition in its ability to traverse the Fibre Channel SAN.
Accessibility to new users needs improvement.
I would like to see better provisioning of storage from PowerMax, as well as automation. If you have a cluster with 15 datastores you have to back and select everything for report groups, as an example. You have to create 15 datastores and bundle them in a storage group. If there was some automation of this process, a script or a workflow, to configure everything, that would be great. Unisphere could be improved and I would also like to see a real-time, graphical view of metrics. I don't know how far back in time we can look, but if we could see the performance from two months or three months back, and how it is performing now, that would be helpful.
I'd like to see the dedup and compression improve. Two to one is not very good. We should be getting something like three, four, or five to one.
I would like to see continued visibility and analytics in the platform.
The NVMe integration could be improved.
We are facing challenges with the SRDF replication which we got with the latest, new box. Overall, it has been working for us so well for the past few years. We hope that whatever bugs there are right now in the PowerMax are going to be sorted out quickly. There is room for improvement with the enclosures. We were happier with the old VMAX enclosures which looked nicer in our data centers. The new one is dull.
We would like easier DR setup. We require DR operations, and we wish it was a lot easier to do that configuration. It could use additional features and easier integration.
There are definitely some improvements that can be made to the CloudIQ. CLI and Solutions Enabler need improvement. While it's been around forever and it's proven that it works, it can be cumbersome to operate and train some storage admins on it. Also, if you haven't been using it for a while, it can be complicated.
There are some stability issues that we just recently experienced. We hope the next release will solve these problems. We probably needed more time to test the solution.
I would like it to support NVMe over Fabrics, because that is the next item for consideration on the NVMe roadmap. PowerMax supports NVMe on the back-end, but when it starts supporting NVMe over Fibre Channel, suddenly various hosts can directly communicate with PowerMax, and with NVMe-oF, as well. Suddenly, Gen 6 and Gen 7 switchers will be able to help facilitate that particular communication channel.
The REST API needs improvement. We are a very big automation company, and this would be big for us.
I would like NVMe to be end-to-end in the next release. Right now, it is not end-to-end.
Our operations team would say that the GUI needs improvement. You taking an island-of-storage and just creating another.
There are glitches in the system at this point in time. I don't get the information that we've traditionally gotten, things like the power report that used to be standard for all EMC components across the board. Now, you can't get them. You get a little bit here in that report and a little bit in another report, but you never get the total amount in one report which gives you the equipment, its power utilization, maximum recharge, the interfaces for the power, and what are the requirements for the interfaces on the other end, so you can know exactly what has to be connected at that point.
There is some room to grow, especially with some of the installation quirks.
I would like to see the rack change. They have defaulted to the standard rack, so our fiber cables are crowded when we shut our back door.
Which is better?