HPE Apollo's after-sales support and technical support should be improved. The solution should provide fast delivery when it comes to replacing a particular product that is under warranty.

I would want to see the flexibility of being able to run various network protocols including InfiniBand, Fibre Channel, as well as iSCSI, with iSCSI going up to 100 gigabytes per second -that would be outstanding. That in conjunction with what Mellanox offers would provide us with a very high-speed networking interface.

The other thing is we may could, perhaps, use more GPUs in the future, go from eight to 16 GPUs per instance. That could run head-to-head against the DGX-1, the DGX-2 that NVIDIA has developed in their own chassis. That would be interesting to see.

In terms of improvement, we could benefit from some training. If there could be some training for our team, we would be better equipped to promote and support HPE Apollo. Currently, we haven't received any specific training related to HPE Apollo throughout my career. If HPE Apollo provides training, we can share the knowledge with our partners, which could lead to more opportunities for HPE Apollo in the market. Deal protection and other support measures would also be beneficial, as they could help us achieve better results in this market.

The solution's deployment, security, and scalability need improvement. 

In addition, a while back, we needed to get a GPU-intensive HPC workload. Our client still needed to accept HPE Apollo because it does not support HPC and that was what the client required. Also, one of our telco customers has said that after the installation of HPE Apollo for their telco requirement, the generically written things regarding the performance were not approved in their operational environment.

One drawback which I had: When I needed to expand storage on the Apollo, I had significant problems getting disks for it. It was a very long wait-time. So, if I were to give any advice in regards to improving this product, I would say make more of the 8TB disks available quicker. I ended up having a few issues because I ran out of space. There was a huge lead time while I waited for new disks to arrive here. It left me a bit exposed there for awhile.

But that's the only criticism. Other than that, I think it's a great product. It's really good. Really reliable. Very cleverly designed and I can't think of what better way they could pack more disks into such a small space, so all around it's a good product.

I think it's on a good track. What's coming out in Gen 10 is very strong in terms of additional security. Overall, I think those are well architected. They're a very flexible form factor for scale-out. Assuming ongoing support for the latest generation CPUs and accelerators, that will be something we'll keep following for the foreseeable future.

In Bridges we combine the different node types to create a heterogeneous, ideal system. Rather than wishing we had more features in a given node type, we integrate different types. We choose different products from the spectrum of HPE offerings to serve those needs optimally, rather than trying to push any given node in a direction it doesn't belong.

It's a very good system when you need a lot of disk capacity. But it's unclear whether the performance of the IO will be sufficient when calculating the theoretical amount of time to read all the disc space. If the workload is not purely sequential, then performance in the IO is less than optimal because it's optimized for streaming processing.

One thing that we are looking for is the better stability of the Lustre file system, it could be improved. I have heard that they are coming out with a better memory bandwidth, so that's good or maybe, it's already there in System 10.

In that case, of course, then there is need for more CPUs, more storage and all of that.

We're pretty happy with the Apollo line of servers. It would be interesting to see the new hyper-converged DL380s. It would be cool to see if that type of same thinking about hyper-convergence was applied to the Apollo line of servers as well. It would be interesting, not on the storage-dense model of Apollo servers but on the compute-dense models of Apollo servers, to see kind of a hyper-converged solution running in those chassis that can have multiple compute nodes all in one. So that would be interesting to see if HPE could do something like that. It would make a compelling argument for them in their hyper-converged space. It would really complement the DL380 hyper-converged solution that they're providing now and would be I think a good choice for lots of people who are looking at hyper-converged.

We are quite happy with it, but its price and storage density can be better.

The solution could improve the hardware, such as the motherboard or servers. We have had hardware faults in the past.

We're looking for faster capability to write to drives. We're fully loaded with all the small form factor drives loaded into the system. It is practically at the limit of the capability supported by the architecture. So we need new solutions, new types of drive capabilities, and faster bus speeds.

We unfortunately have tried to used standardization using Ubuntu Linux and it's been hard. They had some difficulties getting the RAID configuration up and running because there are no drivers for it. It's not supported by HPE.

The predictive analysis feature could be improved. We would like the solution to be able to implement automatically in the next release. 

Obviously I would like to see the cost go down. That speaks for itself. 

We would like to see improved cooling because that is quite an issue. If you put that much compute power into a single rack, cooling really becomes an issue. And there is room for improvement there.

There should be truly independent nodes for your rack, which can contain three different servers. I like to make sure when a component fails, I don't have to take down all three nodes. This is especially true as we usually have replication between these nodes. It would be a great asset to be able to contain the downtime to one of the nodes.

Connectivity to the outside of the server needs to be improved at the same time the network is improving. This would give us more IO. Of course, this is a firmware lifecycle management issue; there is work to do. Vendors should test the firmware before they are delivered to customers.

OpenView is a new product which does not support older versions of the hardware. This is an issue. That's why we cannot switch to the newer one. We continue using the older product, and that's working fine. I would like to see a bit more integration. This is the major topic.

The licensing could be greatly improved, I think. We have a very hard time tracking it because we have to get a license for every server and machine. We have to click in our email, then go to the site, then login to HP, then download the license, then we have to do it all again for each server and machine, and we have to know which server or machine the license is for and give the license to the installer. It's inefficient, overly complicated, and should be simpler and pain free.

We would like to see SimpliVity on top of the Apollo.

Direct integration with software (Cleversafe, Scality, Ceph) for a purpose-built object store appliance. Stay closer to the current rev of processors. I know it is a heating/cooling issue, but being a couple of revs back is problematic when comparing consolidation of workloads with standard intenl servers running the latest chips.