What needs improvement with Cisco HyperFlex HX-Series?

The product doesn’t exist anymore. It would have been nice if the product provided some cloud integrations. That is why the vendor discontinued the tool.

The product does not have a cloud version. I would like Cisco to introduce a cloud version of the solution in the future. The product should be made more flexible in terms of integration capabilities.

The solution's price could be affordable.

Cisco HyperFlex HX-Series could be improved by reducing the number of nodes. Currently, the architecture separates data, computing, memory, and storage into different parts. For example, if I need around 200 servers with a requirement of TB of storage, I would need 200 clusters, which can be quite cumbersome. Other solutions like Dell and Nutanix have a single combined unit that includes computing, memory, and storage, which can reduce the number of nodes required and also reduce licensing and power consumption. Therefore, I suggest improving the architecture of the Cisco HyperFlex HX-Series by combining computing, memory, and storage into a single unit, making it more efficient and easier to manage. It will also reduce the number of nodes required and licensing costs and help to reduce power consumption. I would like to see improved internal integration capabilities in the next release. Currently, it is constrained only to VMware only. So, there is no integration part of the RAC or IEL or anything, and it isn't easy to manage it. If I want to hold another partner, then that part will not be possible. So, you have to manage that particular integration. Another feature could be in terms of memory usage.

We'd like the product to offer better integration with other products.

In our case as a credit recovery company working with the state, our infrastructure is fixed and stable as we don't have the same need for growth as other types of companies. That said, one of the biggest problems with HyperFlex HX is that if you want to adjust your solution in terms of processing power, memory, or disk capacity, you have to buy completely new hosts. From a financial perspective, it can be very expensive to do so, and from a legal perspective, there are all kinds of compliance issues we would have to sort out before buying any new solution or application. Since we as a company will only be operating for perhaps the next ten years, when our HyperFlex solution comes to its end-of-life in 2024 our next step will likely not be with HyperFlex or any other solution from Cisco. Instead, from a financial point of view, we will likely turn toward a cloud solution because, that way, we won't have to spend so much money on physical infrastructure. We are, after all, only a small company and HyperFlex can be very expensive for other companies of our size, whether in Portugal or elsewhere like the United States. One other area for improvement is in regard to HyperFlex's integration with VMware. HyperFlex integrates with a specific version of VMware in such a way that HyperFlex doesn't always resolve security issues with VMware at the same pace as what you see in native implementations of VMware. This has happened in one or two situations in the past. On the other hand, the integration is otherwise adequate, especially in terms of availability and virtualization features (such as being able to split up each of our four hosts within VMware).

HyperFlex, compared to other competitors like VxRail or Nutanix, has stability issues. The utilization needs to be better. It needs more options. For HX, we need to install a specific virtual machine on each node as a controller. For VxRail, we don't have to do this since it's a built-in feature on VMware. For the solution to work, we need to buy specific switches from Cisco. It's not flexible, actually. I wouldn't recommend using it. The cost is pretty high.

A lot needs to be improved. Cisco HyperFlex HX-Series does not support the stretch cluster, and the interface is not good. Features are missing, and it does not support VMware Seven.

The initial setup could be faster, due to the prerequisites. Having a small infrastructure that includes Nexus and fabric interconnects is costly, which is something that should be improved.

I'd like to see easier integration with other vendors. For example, it wasn't that easy to integrate with VMware initially but eventually it worked. Ideally, I'd like to see the option of integrating with those devices on cloud, to use storage from our on-prem in a bucket in the cloud. We haven't yet seen any documentation regarding that. We need to find a way to share resources between on-prem and cloud.

Every node has a storage controller VM that checks identity and I think that should be eliminated in order to optimize the resource utilized. I'd like to see the ability to integrate with HPE because SimpliVity has some backup-live features which are useful for us. Finally, some improvement to the documentation would be helpful.

We had some issues with deployment, specifically with Intersight and deploying HyperFlex on-prem. It was difficult to figure that out with support so we spent a lot of time with them in order to get the solution deployed. That would be the main area where I'd like to see improvement. For a future release, I'd like to see the Containers Platform added to the product.

The solution could improve by having more customization.

The price could be better. I think it's more expensive than VxRail. In the next release, I would like to see a better level of protection and availability with six clients.

The problem is that we always have to keep an eye out for new releases to make sure that the version we're running is up-to-date. The way their infrastructure is updated and evaluated, they are always updating their firmware structure. That is the only challenge we are experiencing, but you have to look for the updates all the time, which is a challenge and everyone's very frustrated at this point. The cloud world is changing every day. It would be good if they added provisions for activities like micro-services and micro-segmentation.

If you need to expand, scalability is an issue. Overall, you can just call Cisco and then they will help you to fix everything. The storage could be better. If you want to expand the storage, sometimes you need to expand the memory at the same time. This could be cumbersome or pricey. The initial setup can be complex.

Cisco HyperFlex HX-Series could improve by having the documents be more readily available and the support could be more flexible.

They need to provide lower prices. It is pricey in comparison with other vendors - including Nutanix, or VxRail of EMC. They are quite expensive due to the fabric interconnect, the FI, that they have. The other vendors don't have those fabric interconnects while Cisco requires the fabric interconnect. The initial setup can be complex. We'd like to see a backup solution included in the product in the future.

HyperFlex could be improved by reducing the minimum number of nodes supported from three to two. In the next release, Cisco should include full integration with other virtualization vendors like Nutanix, the same way they do with VMware.

You have to get the same servers with the same storage; they need to be identical. However, in vSAN or in VMware we don't have to do that. We can just add storage and manage it in the same server. In the next release, I would like to see more integration with VMware. I would also like to see more automation with other Cisco products so that we could have one dashboard for all Cisco products.

I have been receiving a memory overflow message every day when I use the solution. This memory overflow error is a cosmetic error. Over the last three years, I have been talking with Cisco to remove this error. Cisco and VMware should cross-check and troubleshoot this issue to come to a solution. They should remove it as soon as possible.

In terms of what could be improved, Cisco HyperFlex should decrease the amount of memory needed from the Controller VM that controls the physical discs. They control the discs by using the virtual VM over every ESXi host and the VM consumes memory and consumes more hardware resources. They have to improve that by decreasing the amount of required memory and CPUs to control this disc on the server. From a technology perspective there is no problem. The main issue is cost-wise. Cisco costs more than its competitors. As for what I would like to see in the next release, Cisco already has a solution, but in a different profile called the CWOM, Cisco Workload Optimization Manager, and Cisco Intersight. This solution manages and optimizes workloads on the infrastructure and integrates them with the monitoring solution and gives insights and recommendations for the customers. I think it is included on solutions such as the VxRail which has the vROps, vRealize Operations Manager. Cisco did not include this on this solution, but in a separate solution. They have to have a logs server to collect all logs and give insights from all hardware. This is included in the Cisco Intersight but it is not included in the hyper-converged infrastructure solution.

The pricing can always be better on everything. The interdependencies of each of these functions and the configuration side are something that needs to be carefully architected so that if something isn't working in my memory stick, it doesn't have a cascading waterfall effect through the rest of my operation. The interdependency could be improved so that everything will not be so interrelated. The initial setup can be a bit complex.

In the next release, Cisco should add more integration and management capabilities as well as some tweaks to the dashboard that make it more user-friendly. They could also add support for multiple hypervisors.

There should be the opportunity to create more than one div group. The solution has only a gigabyte boot device, boot drive of only 240 gigabytes, but 480 would be better. It is possible to create two or more div groups, for example: having two test drives and five capacity drives for each div group.

The minimum for running involves three nodes, which is the number I use for running the patch. The solution is not very good when it comes to a single node. 64 is the maximum number of nodes. As I work with Hypervisor or VMware, the solution usually is easy when it comes to monitoring and backup. I rely on backup and monitoring capabilities. I use Veeam, for example, as a backup solution and SolarWinds for my monitoring needs. I am a gold partner of Cisco and am entitled to a discounted price. When it comes to customer needs, the licensing price could be better. By comparison, Nutanix is more affordable. The choice of Cisco varies with the needs of the client.

A disadvantage is the higher costs involved in using this product. If it were more affordable, it would be easier to recommend and HCI adoption rate would increase. Unlike other options, you need to pay a subscription to Cisco yearly instead of paying for the hardware outright, which makes it more expensive in the long run. The initial setup could be easier. Right now, it's a bit difficult.

They need to make many improvements to this solution but the most important area is the compression. Most customers are concerned with the compression for a specific workload, and then maintaining it. The performance and compression vary depending on the type of workload; for example with SAS HANA, Cisco forces compression and that affects performance very badly. At the least, the customer should have the option to choose what types of workload should be affected by the data reduction functionality. The second point is that they need to work on the erasure code. Cisco doesn't support erasure code, even over flash. If they fix this then they will only waste 25%. With replication and mirroring, it uses 100% of the allocated capacity. This means that Cisco needs to work on the architecture. I have conducted many PoCs and it is a problem that they need to work on. Cisco offers a single cache drive cluster, whereas VxRail offers up to four cache disks per drive. When the cache disk fails in Cisco, the whole node fails, and the workload goes down. But when it comes to VxRail, if a cache disk files then only the node fails and the workload remains up and running. This is an area that Cisco needs to work on. Essentially, they have to raise the number of cache disks that can be included in a single node.

There could be an increase in performance to improve the solution.

The pricing of the solution could be improved. It's a bit too high. We find that most customers can not make it work with their budgets. When it comes to the virtualization layer, we have issues. We can scale up, and yet we can't add storage. We need to be able to scale out and not just up. When you want to scale up or scale out, you are quite limited. It would be ideal to have the flexibility to scale out whereby you are able to do a single type of commodity upgrade. For example, if we could upgrade the memory only, or upgrade the process only or storage only,. There is a general over-reliance on VMware, the form of the software layer which now includes things like Nutanix. Some customers - especially around Oracle - have a preference to not work around VMware. That's why we need more flexibility to be able to do any event on the top layer. It would be ideal if we had local support here in Kenya.

This solution is lacking in replication and backup abilities that I would like to see in a future release similar to HPE SimpliVity.

If we compare it with VMware, the VMware product is basically very open, and it can be easily integrated with any platform. VMware product is also available on the cloud and is not an appliance-based product, whereas Cisco HyperFlex is an appliance-based product. Companies that want to use HCI as a platform or as a service would prefer something without an appliance. A SaaS-based product is better for a customer who is using the cloud and has multiple resources, workflows, and devices on the cloud and wants to go for a hyper-converged solution. This is where improvement is needed. In addition to an appliance-based product, Cisco should provide a SaaS-based product. Its price should be lower. Cisco needs to work on the pricing model for this product. Its price is a big road blocker when competing against Nutanix and VMware, which are its main competitors.

There are sometimes issues with memory failure. I'd also like to be notified when new documentation comes out, it would make things easier. I think the cost is an issue given that it's so much cheaper to use cloud based solutions these days. I don't know whether HyperFlex can compete with that. There can sometimes be issues if you change your mind on something and want to move things around.

As a fairly mature product, I can't think of any features that are lacking per se. The pricing is pretty expensive. Right now they are coming into the public cloud, we want to be seeing HyperFlex integration with public clouds such as AWS, Azure or Google. I need to see the feature more use cases for the backup and restore functionalities or how we can expand the cluster to the public cloud.

There are a lot of features that could be improved. We sorted out one critical issue left by the initial deployment team, our external partners. They made a mistake, a wrong configuration regarding the network settings. That left us a very, very painful troubleshooting process. Eventually, after four years, we sorted it out. After fixing that the system was stable. We even thought that maybe this is the root cause of those hardware failures. In the next release, I would like to see them able to connect to the public cloud.

Its price could be better.

The scalability could use improvement.

We would like to have the ability to not have to reboot while doing updates. Being able to work through updates with as minimal amount of impact to users.

Having this solution even easier to manage would be nice. The deployment could be made a little bit easier, as we still seem to struggle with it a little bit for a day or so to get it running. I would like to have less clicking to deploy it, where it is ready to go out of the box with a default build. That would make things a little faster.

They should fix the upgrade process. The upgrade process is great. It is seamless 70% of the time, but when it doesn't work, it goes south. Cisco needs to fix some of those upgrade bugs and it will be a perfect solution. That would be good for somebody who doesn't want to spend a lot of time micromanaging the device in production.

The additional feature I would like to see included in the next release of this solution is more security. We want to add more data servers to all the hosts all over the world. We have 140 hosts that connect to one data center in Manila.

We would like to see better integration in the next release of this solution. Managing this solution is complex and should be improved.

I would like to see more flexibility in the storage. If you have experience with Cisco, or you have experience with the HyperFlex HX itself, then it's a lot easier. If you do deployment on your own, there's a pretty steep learning curve.

Maybe a better version is required to gather all of the information to allow HyperFlex to bring everything online more rapidly. The initial implementation workflow could be improved. There's a lot of information you need to gather. At the time we deployed it, approximately a year and a half ago, that there was a lot of information that we needed to gather. It wasn't explicitly called out to in the workflow as you were going through the screens. We had to know what we were doing. We read a lot of documentation to get all of that information.

One problem is that whenever we want to do something on VMware, you are dependent on HyperFlex. With everything now, you have to look into HyperFlex. You cannot just apply the VSXi updates. You are dependent on Cisco HyperFlex for metric services, analytics, and everything else. I don't like that dependency.

I don't work day-to-day enough with Cisco HyperFlex to be able to say what would necessarily make it better. Initial integration was tough. Our group was just looking for another approach to hosting their various VMs.

The infrastructure team thinks that there are a lot of issues.

I think the product can improve dramatically in the area of deduplication. With deduplication as it is, you have 30 or 40% of the storage that can't be used because the deduplication is not efficient enough. I hope they fix that problem.

One of the challenges we have with Cisco, in general, is the multiple management platforms for the UCS, and I would like to see it all under one umbrella. If we could have just a single pane of glass, where everything can be managed through that one app, rather than have multiple apps, it would be great. There are three apps that we currently have to integrate with. I would like to see them broaden their processor options. We've had some challenges with Intel, and I'd like to see them look at other potential CPU providers.

I don't really think it needs to be improved as far as the product itself. I can't think of anything yet that it doesn't already do. It fills our needs. It would be best if it continued to meet the needs of technology as it advances over time. I think the documentation is a big con for Hyperflex. More substantial, straightforward documentation would help end users make more of their investment. There really isn't a problem with product functionality.

One aspect that really needs to be looked at, is to fix the bugs issue.

Single console management (like Hyperflex and FIs management). I mean to say that, we need to log in 2 console one for vcenter and another for Hyperflex manager, this should be on same console page either hyperfex manager settings and monitoring in vcenter or vcenter can be managed by hyperflex console too.

It needs improvements in terms of I/O. Sometimes I have felt it is slightly slow while we are trying to provision VMs or access VMs. They should concentrate on that issue. At the same, time L2 switches are bundled in HyperFlex. If they make that the L3 switch, that would enhance the performance slightly better.

Cisco has to continue improve the management tools to provide a better command line interface with more functions. That would be better for administration.

I would like it to be easier to manage the server, create VMs, and manage the VMs on the client.

I would have liked it to be able to back up the system to a cloud, to be able to work from any other place. If there is a data disaster, you need somewhere where the data is stored so the company can still work, especially if you have several other facilities working on the system remotely.

We would like to move off the GUI.

They should improve the price. A lot of people complain that the price is an issue. For smaller countries, this solution is too expensive.

We would like HyperFlex to connect the storage directly into Fabric Interconnect along with the features of the solution. We have enough ports in HyperFlex. We don't want to buy another set of switches to connect to Fabric Interconnect, complicating the solution. This change would be a great improvement.

They should have more blade options and more configuration flexibility. Also, the price is expensive. It should be cheaper but it is worth the price.

We are moving further out with more cloud solutions, so we need HyperFlex to be more cloud-connected. They should develop more connection to Amazon Web Services and Google Cloud because our customers use their services and are moving in that direction.

The primary improvements should be made in the cluster storage controller VMs so that they don't break the upgrade process. I would like to see integration with multi-site ACI deployments so that we can span clusters between multiple sites, with ACI running within each site.

We are working with a lot of different technologies and trying to manage them with one tool. The API could be improved for better integration with other systems. For example, the integration to VMware could be better.

Cisco should improve on having more technical documents available on their website, as well as keep them update-to-date. I would especially like to see more compute documentation.

There is a VMware plugin for HyperFlex, which sometimes it hangs up in our environment, and doesn't function well.

I would like for them to develop integration with other technologies, like SD Access. It's still lacking an enhancement management player.

I would like to see them expand onto the cloud.

They can't get a POC to work because there are big latencies between the two sites we have. Even though we have a private fiber connection and everything else works well, they can't get the two products to talk to each other. They are not able to build it up. It looks like it's a firmware issue. They said it isn't but now it's looking like it is. I hope we can sort it out.

Mostly documentation needs to be improved, we need strong documentation. We also want to not need to open a tech case every time we need to troubleshoot. Stability and storage need to be improved.

I would like to see more analytics. The storage data is just IP packets. It could use better infographs in the HyperFlex Connect on how traffic is running in the network. If you were reaching any capacity issues on the Fabric Interconnects, it should be able to cool all of the servers and Fabric Interconnects, then possibly integrate it with, e.g., Nexus Series switches. This should all be available in a single pane of glass. I would also like a fast on-premise service. While there is Intersight, which does a lot, there is not the same clarity of information from Intersight, as with an on-premise service.

Cisco should combine its features with its prices.

A stretch function is really needed. Germans think that because of the high availability and the security that two data centers are better but we want edge all in one cluster. This is a needed feature, the cluster feature, and we'd also like to see a possibility for ROBO offices.

The product's disadvantage is that it is not necessarily open as I would like it to be. The main platform has some closed features that I would rather not see. For example, I need a mechanism to orchestrate services and an infrastructure as a service model, which is how we are deploying it. We would like to orchestrate it using an open, non-Cisco tool. While there is an API, in order to deploy it, we need to indicate it with a higher layer Network Services Orchestrator. This requires a lot of work to be done. If it could support a more open model, where we have standardized blueprints, templates, or virtual machines, it would be nice. This is an industry problem, not necessarily specific to Cisco. I would like a lighter product; something which can be carried by a mere human. The typical size still looks like a stack of data center-sized servers, which is typically fine, but not so good if you want to carry it around.

Some of our customers do not want to adopt the Cisco ACI solution. They want to manage Cisco HyperFlex by integrating it with their own management solution.

I would like to have an easy solution calculation to plan my infrastructure and complete my licenses. Because as of now, I have to read a lot of materials and documents.

I would like to see the capacity to choose the replications, RFS sectors to all machines, not only on the cluster. Additionally, the solution should provide more detailed documentation.

Deployment scripts can be improved. Several clans need to be created before the deployment effectively works.

We are operating two clusters for site resiliency. I would like to support running redundant/resilient vCenter server instances on each rather than a supported external server.