What needs improvement with Dell EMC PowerMax NVMe?

There is still room for improvement in the architecture and scale-out structure of the platform. While Dell PowerMax supports data reduction and intelligent data placement, its scale-out growth model could be further enhanced. Currently, PowerMax can scale from two nodes to 16 nodes, expanding from terabytes to petabytes based on workload and application requirements, but greater flexibility could reduce dependency on pre-planning. Additionally, improvements could be made in areas such as AI integration, user interface, support, and customizations. From a performance perspective, PowerMax has evolved from the 2500 and 8500 series, which provide enhanced features for multi-tenancy, container workloads, multi-access protocols, and channel connectivity. These models also deliver better security, higher performance, and greater capacity, though continuous refinement is still possible.

There are quite a few design areas where I have been trying to understand why it has been designed in certain ways. This includes redundancy at the engine level, the virtual engines, or the directors that are installed. It is also related to how the cache destaging happens, or how it's designed, and how read and write caches are designed. These are design areas where I think they can do better.

Performance and management aspects are satisfactory, however, there is room to enhance data protection and future proofing to enable remote visualization and control capabilities when compared with other vendors. It can improve its hybrid model. Improvements can be made in support services, remote support, and remote visualization. This depends on in-situ support, remote support, and other factors, however, enhancement is needed in our territories to improve remote support and parts availability. While Dell PowerMax NVMe is very reliable storage and cannot experience downtime, we sometimes face challenges in our territories regarding the availability of critical parts. Our local support people occasionally encounter situations where faulty parts are not currently available.

Dell PowerMax or Symmetrix is a block-level storage and unified storage, supporting both NAS and SAN. One area for improvement would be integration with public cloud services. Many users are inclined towards hybrid cloud solutions, but it remains primarily an on-premises solution. Compared to competitors such as NetApp, which integrates with public cloud hyperscalers (GCP, Azure, AWS), Dell PowerMax lacks in this aspect. Integration with a public cloud would provide an added advantage and give it an edge over other vendors.

The disadvantage of PowerMax is its cost. It is expensive, mainly due to pricing. The issue is only with pricing, as I can find the PowerVault ME cheaper than PowerMax.

Dell PowerMax needs more connectivity options for projects that require separate web access. Usually, users propose PowerMax, however, this could be better addressed. Additionally, features related to AI technology could be improved by learning more from Pure Storage.

There is room for improvement in adjusting into the existing infrastructure easily without much of the technical contribution.

The tool's disadvantages are its cost and lack of ease of use. According to the latest market trends, certain technologies are lacking. The main challenge with the product was to have control over the simultaneous failures of controllers. For recovery, manual intervention is required. The tool's cost, active controller part, installation, maintenance, scalability, and recovery parts can be considered for improvement.

The solution should integrate better with OpenShift and Kubernetes architecture. We also use OpenShift in our environment, and I had some problems integrating Dell PowerMax NVMe with the OpenShift Kubernetes environment. The solution's integrations and usability could be better.

I would like to see more security features and improvements in the platform's ease of management. Sometimes, it isn't straightforward to understand, especially for new users.For example, if there are power issues and the system needs to be shut down, we need to connect with the support team and go through several steps. There are complications to powering down the system. The backup features need enhancement.

Configuration could be easier to manage, and implementation should be easier to install and configure. Dell can improve the Service Level Agreement (SLA) compliance. Sometimes, when we need to replace a component, the SLA says four hours, but for some reason, the technician doesn't arrive until the next business day or even six to eight hours later before the case is placed. So, focusing on respecting the SLA is where Dell could improve.

The initial setup process is difficult.

The price cap of Dell PowerMax NVMe is very high. In Dell's portfolio, there are PowerMax and PowerStore, and it is important to note that there is too much of a price gap between both. The aforementioned aspect of the solution related to the pricing element is an area where improvements are required.

The solution does not use new versions of OS and patches. Its installation is also difficult. The solution is not as fast as other storage in the market.

Dell PowerMax NVMe needs to increase storage and improve scalability.

Dell PowerMax NVMe is costly compared to other solutions.

The tool needs to improve its performance. Today's applications are demanding a response rate of one millisecond or below. The product should also look into AI integration.

Some features could be better. I think if we can integrate PowerMax with the public cloud, it would be safer as we can have the service without any data on-premises or in the public or hybrid cloud. And we can remove all the data from on-premises and the public cloud. This would make migration and data recovery quicker. There is room for improvement in terms of integration with various service providers for public clouds. It would be beneficial to integrate with AWS, Azure, Microsoft, Oracle Cloud, and other service providers. This integration would enable a more comprehensive and open solution, catering to multiple vendors and applications, particularly in Azure cloud. This would enhance the overall customer experience and provide more flexibility.

The tool is costly compared to other similar products. The product's pricing needs to be improved. I would like the product to include the replication feature in its future releases.

Dell should work on their marketing strategies to make the product more visible on the market. They should promote the product's compatibility with IBM, as not everyone knows it.

The solution is quite expensive and I believe Dell should examine their prices because they are currently very, very high.

We would like to improve the delivery model. They don't have the delivery terms today. If they improve this area, everybody would be happy. For the Romanian and Eastern European markets, we have long delivery times, and even if we have a solution, we don't have the products, and we don't know what to install. For the time being, manufacturers must enhance the delivery of all items while they have them. If you go to Pure or you go to NetApp, they deliver in one month. If you go to Dell or HP, they will take up to six months for delivery. This is not a business model for today.

Simpler management would probably be my biggest ask. Some of the management features could be simplified and that's probably the main thing they need to address.

I would like a more informative CloudIQ for iOS. What you can see via the web UI significantly differs from what you can see via the web application.

The CloudIQ features still need to be improved because CloudIQdoes not support PowerProtect DD capacity, although it is working well overall. Their mobile app also still needs improvement. In addition, the web GUI is good and shows all the related reports, but I would like to see more granularity in the reports, and reporting on CloudIQ should be done in the web GUI interface. There is also room for improvement in the PowerMax architecture and hardware itself. They should design the PowerMax on the basis of PCIe 4.0. I would like to see the possibility of an NVMe drive that operates on PCIe 4.0 and not PCIe 3.0. The design could be very much better if they did some R&D and introduced a version based on PCIe 4.0.

The dedupe and compression features have been the biggest disappointment. It's not as efficient as we were expecting or had hoped. It's not terrible, but not as good as we were led to believe it was going to be. They need to improve their reduplication algorithm or the compression algorithms. It comes with a guarantee that you'll get 3-to-1 dedupe and compression, meaning that if you have 3 terabytes of data, it should only take 1 terabyte of space because we reduce its size. We're only getting 2-to-1. It's not a big deal because we have more storage than we'll need, but it's disappointing. There's also a qualifier in that I'm told that if we filled the array up more, some deeper algorithms would kick in and help that reduction number go up a little. Also, if you have deeper algorithms that you're going to use, only if I put more data on it, is that going to slow things down? Why not just use them now? That also left a lot to be desired. I attempted to use that and was having some performance issues, and the fix was, "Don't use that." So it was a little lacking. I think management is where PowerMax is weakest. We're still managing it like we managed EMC arrays in the early 2000s. There's a slicker, fancier GUI that does more, but at the end of the day, you still have to dig into the command line and issue a lot of the same commands that we still were using almost 20 years ago. So the ease of use factor is low. One of the reasons I wanted Pure Storage was because I felt like I could teach a coworker how to fill in for me if I ever went on vacation for a couple of weeks. If anything bad happens and I'm out of the office, they're going to have to bother me. This is not intuitive. There are a lot of CLI commands that you still have to use. It's just not as user-friendly as it should be.

They can make the GUI better, especially for the ones that come out of the box. We did encounter a bit of difficulty in setting up the storage. We had to deploy Solutions Enabler on a Linux machine to be able to fully interact with the storage. They need to upgrade the web interface for the management of the storage that comes out of the box. The management interface for NFS is also a bit old and not very intuitive.

We would like more documentation, a guide to the features of the PowerMax. We needed to use the option to reclaim storage and we had to chat with the Dell EMC team.

We brought up this question to the implementation engineer. We were comparing use cases where a customer is using RecoverPoint, then goes to PowerMax. In our previous setup with XtremIO, we were using RecpverPoint and keeping snapshots for 30 days, every few seconds. With PowerMax, I requested this for every 15 minutes, keeping it for a week. The engineer's answer was, "There will be too many snapshots. It might slow down the system." This is specifically for the use cases where there is RecoverPoint. While PowerMax works with RecoverPoint, and you can use it, there should be some way where you can have even more snapshots and not to worry about performance and system cache.

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

It would be nice if there was a training course offered by Dell EMC. It would help us use the product.

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.

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.

There is some room to grow, especially with some of the installation quirks.

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.

Our operations team would say that the GUI needs improvement. You taking an island-of-storage and just creating another.

I would like NVMe to be end-to-end in the next release. Right now, it is not end-to-end.

The REST API needs improvement. We are a very big automation company, and this would be big for us.

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.

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.

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.

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.

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.

The NVMe integration could be improved.

I would like to see continued visibility and analytics in the platform.

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.

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.

Accessibility to new users needs improvement.

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.

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.

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.

There is room for improvement in the replication. It's an important requirement for us.

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.