Atlantis USX Review

There are a lot of different types of ways to deploy this: Hybrid (SAN/NAS or Local Storage), Hybrid using DAS, All-Flash, In-Memory, Simple Hybrid (requires only one host), Simple In-Memory.

Which features have you found most valuable? Let's start here... We've found that Atlantis USX helps remove some major inefficiencies around storage silos by unifying all storage types into a highly-optimized pool of storage resources. These resources are then made available to virtual workloads, desktops, and applications. From there, you can control these resources (again, across a number of different types of storage repositories) from a policy-based perspective. This allows you to optimize capacity, performance, and overall storage management. There are a couple of big features that we like with Atlantis USX. First of all - there are a lot of different types of ways to deploy this: Hybrid (SAN/NAS or Local Storage), Hybrid using DAS, All-Flash, In-Memory, Simple Hybrid (requires only one host), Simple In-Memory. Now, aside from all of the other features you can wrap around your storage repositories (deduplication/compression, caching coalescing, encryption/data protection, HA/DR, remote replication) - you can also use USX as a vVOL engine. This has been really useful. You can take legacy storage controllers and give them new life again by pointing the resources to the USX virtual controller. This means that the underlying storage system does NOT have to be vVOL-enabled; rather the USX controller is the engine for that. Cool stuff.

Any examples of how it has improved the way your organization functions? Storage has been a challenge for a lot of organizations. Our customers are constantly looking for ways to optimize storage systems, re-provision legacy gear, and improve user experiences. The flexibility behind USX allows it to integrate with a few types of hypervisors (XenServer and VMware) and control numerous storage types. From there, using REST APIs, you can provision storage resources between on premise and even remote (or cloud) locations. What's cool is that heavily virtualized organizations can integrate with vCenter, vCAC, OpenStack, CloudStack and even IBM Cloud systems. The major improvements really revolve around the amount of storage control an organization can leverage. No more lost resources, no more silos, and no more fragmented management. 

What are the areas for improvement? There are a lot of things happening "under the hood" when working with USX. There's a lot of scripting, orchestrating, and deployment automation that goes on where the admin doesn't really have to worry. However, pay particular attention to how much RAM your hosts have. This can limit your ability to create volumes; or sizing them. Capacity and resource planning is critical here; otherwise you run the risk of performance degradation. One more thing to keep in mind - because of the tight coupling between hypervisors like VMware - you have to be aware of specific sizing metrics. For example, if you're running VMware 5.5, the default queue depth for NFS is 4294967295. To prevent NFS volume disconnects - you should decrease the queue depth to 64.

How long have you used this solution? 2010

Did you encounter any issues with deployment? As mentioned earlier, the most important aspect here is planning, sizing, and proper configuration. Like the example above - if you're going to run Simple Hybrid or Simple In-Memory volumes, you have to adjust the NFS maximum queue depth. Otherwise, you might experience NFS volume disconnects... and that's not fun. Ensure you properly size around RAM requirements; and know the workloads you plan on deploying. 

Did you encounter any issues with stability? Probably the ONLY times we'll see issues with stability is when there are sizing issues or poor capacity planning. For example, if cache and RAM aren't configured properly, VMs might reboot - or you'll have poor performance. Similarly, make sure you know how the USX virtual controller is integrating with your hypervisor and underlying storage. I highly suggest reading the deployment documents to make sure you don't miss anything. If you do the deployment right - the technology is extremely stable.

Did you encounter any issues with scalability? You start with a minimum requirement of 3 servers (they also have a 2U 4 node appliance). But let's focus on the USX virtual controller. I've never run into a scale issue. Beyond the original deployment, you can simply add more severs and scale up by adding more disks to each server node. As opposed to some other solutions, USX actually allows you to combine different types of servers and disk configurations using new or existing servers. From there, you can linearly scale to pretty much as many servers as you can handle and several petabytes of integrated storage. 

How would you rate the level of customer service? Not perfect - but pretty darn good. They've got field people all over the US (and in Europe/APAC) as well. From there, their SEs are very knowledgeable and very friendly. They do a good job integrating with the channel to support sales and technical efforts. 

How would you rate the level of technical support? Their technical support is very, very responsive. They'll get on a phone with a partner or customer very quickly. They'll even get an engineer or architect onsite if needed. Overall, solid with a lot of smart people!

Did you previously use a different solution and if so, why did you switch? It's not that we used a different solution - we just have a few "SDS" platforms that we resell. One of those is VMware vSAN. There are some core differences in these technologies. For customers who have a very heavy VMware infrastructure and not a lot of disk diversity - vSAN can be a great product. However, if we're integrating legacy gear, wanting to use vVOL on older systems, integrating heterogeneous hypervisors, or are trying to create a more customized SDS solution - we work with USX. It's more so around the use-cases rather than "switching" products.

Was the initial setup straightforward or complex and in what ways? The initial setup is probably one of the easier you'll see for an enterprise technology. First of all, so much of the deployment is automated. It's clear they've put in A LOT of effort scripting the deployment and making it as easy as possible. Basically, there are 5 steps, you configure the system (for the most part) via simple click and select, and you go through the install wizard. You can of course customize it as you need it -- but honestly, out of the box - you can have this thing setup in under an hour or so.

Did you implement through a vendor team or an in-house one? If through a vendor team, how would you rate their level of expertise? Well.... we are a vendor and we've done several of these installations. For more complex deployments I always recommend working with a partner who understands your overall infrastructure. This helps minimize errors, deployment challenges, and properly sizes for future use-cases. Finally, even when working with in-house teams, use a partner to ensure you have the install done properly and validate your configurations. 

What is your ROI? Honestly - it'll range. Your cost per server will go down between 20% and even 60%. If you're doing VDI, you'll see the cost per desktop go down as much as 40% to 75%. 

What do you advise others about pricing and licensing? So, licensing is capacity-based. The cool part is that you can do a demo or PoC of the technology and then just add-on capacity licenses. On that note... Plan, plan and plan!! You won't run into any licensing issues if you properly plan out your capacity requirements. 

Before choosing, did you evaluate other options? If so, which ones? As mentioned earlier - we do use vSAN. However, that technology fits specific use-cases. We've seen that USX is a bit more versatile and can fit into more enterprise and more deployments.

Do you have any additional comments or advice regarding this solution? There are a lot of great features within USX that we leverage and see customers leverage as well. Everything from self-healing to automation and event-based management can be configured. This is a powerful products which deploys VERY quickly out of the box. This can also be kind of a bad thing. Please make sure you understand all of the features within the product. Know when to turn on/off deduplication and compression around specific workloads, know when to create replication, ensure that hosts have enough RAM for a volume, and so on. As quick as it is to deploy this, ensure you understand the product and everything that's under the hood. 

Disclosure: My company has a business relationship with this vendor other than being a customer: We are a VAR which resells Atlantis products; along with numerous other solutions as well.
Add a Comment

Sign Up with Email