Intro to PowerStore

To start off the datacenter tech content, I thought I’d start with a fairly new product on the market – the Dell EMC PowerStore. I recently had the fortune to add a 2U PowerStore 3000T appliance to the lab at work to replace our VNXe 3200 that we have had for a number of years. The All-Flash system is a new, mid-range, unified storage system to provide SAN and NAS capabilities. The system is nice and compact while still offering high-end performance and the ability to both scale-up AND scale-out. Working with the Dell EMC professional services team, my team had this up and running to begin testing in a matter of hours.


The PowerStore deployment process was pretty quick and easy once we started. The key to the deployment being successful seemed to be the importance of a properly configured network. Networking best practices requires 4 VLANs for the T-series deployment. A VLAN for Management, Storage, and SDNAS, and an untagged VLAN used for private communications including IPv6 traffic. Within each of those VLANs, you just need to allocate 4x IPs to management communication, 3x IP addresses to the storage network, and 1+ IP addresses for the SDNAS capabilities if you decide to utilize NAS. The PowerStore also utilizes LACP for the connectivity between the I/O Modules and your Top of Rack (ToR) switches for resiliency and load balancing across links.

During the deployment process, you must determine if you are deploying the PowerStore as a block-only storage system or as a unified storage system with block and SDNAS – this cannot be changed after the fact. In most cases, the unified storage systems is my personal recommendation with the only real caveat being that extra compute resources on the PowerStore are consumed to provide the NAS capabilities. The SDNAS deployment also requires the untagged VLAN to pass IPv6 traffic and be allowed on the trunk port.

Standard network layout from PowerStore to 2x ToR Switches and an OOB Management Switch. LACP for the I/O modules and VPC (Cisco Nexus) or VLT (Dell PowerSwitch) between the switches. Tagging multiple VLANs across the links and a single untagged allowed VLAN for intra-cluster management and data (ICM and ICD) traffic.

First Impressions

It took a couple days, but I’ve been able to migrate just a little over 4TB of data to the new PowerStore. I didn’t utilize PowerStore Manager’s external storage migration feature that can automatically import volumes from other Dell storage systems (SC, VNX2, Unity, etc.) and cutover the storage to the new PowerStore volume – testing for this will definitely come in the future. I was able to move about 100GB of files (primarily ISOs for the lab) to the new NAS share on the PowerStore via Robocopy. I then moved about 65 VMs, including Horizon VDI instant clones, on our ESXi cluster to a new datastore cluster provided by the PowerStore.

Our old VNXe had spinning disks and no dedupe/compression capabilities, so this is where the PowerStore impressed me most. PowerStore has a 4:1 data-reduction guarantee (although this has qualifiers as does any data-reduction guarantee), but in my experience with data-reduction, this is something to typically be skeptical about. After moving my lab entirely off the VNXe, my 4TB of consumption was reduced down to only 1TB being consumed on the PowerStore.

The performance in my lab has also been greatly improved. Since migrating everything to the PowerStore, I still have barely crossed the 0.5ms latency mark. I will qualify this by saying that none of the workloads running in the lab are high-end, production quality loads. But… this is still note-worthy in my opinion. True performance testing to see what the PowerStore really is capable will come in the future – stay tuned.

Performance of the lab PowerStore over the past few days – 0.5ms latency max so far during storage migrations.

Upgrade Process

The PowerStore is a solution developed with a container-based software stack meaning updates should come frequently and be very easy to apply. With the initial upgrade applied to our PowerStore after deployment, I saw firsthand how seamless the upgrade process is. The process was primarily a matter of uploading the new software package, performing an automated health check, and starting the upgrade. The upgrade is non-disruptive. It utilizes the active/active node configuration to upgrade one of the two nodes in the appliance at a time. It then performs a per-node rolling reboot to apply the upgrade. The only “downtime” is when the management interface disconnects temporarily while the cluster management IP migrates between nodes.

In less than an hour, I had an upgraded system with zero downtime. This is exactly the kind of capability that I look for when installing a system for a customer and exactly the kind of capability I like to advertise and show off when working with my customers because of how easy it makes their lives.

Room for Improvement

The only area that I felt was lacking from the PowerStore compared to other Dell EMC storage systems such as the Unity family was the level of integration and orchestration abilities with a VMware environment. One of my favorite features of the Dell EMC Unity solution is its tight integration and orchestration when it comes to providing storage for vSphere. With the Unity, creating datastores is as easy as connecting vCenter to scan for the ESXi hosts, configuring iSCSI VMkernel Adapters on your ESXi hosts, and the Unity automatically connects the initiators in a matter of minutes. You then just create datastores from the Unity management interface, Unisphere, and provide access to the hosts.

The PowerStore so far does not natively seem to have the same kind of capabilities. With PowerStore, you must manually register the PowerStore as a VASA storage provider in vCenter, but this does not automatically add your ESXi hosts into PowerStore. It also seems that VMs on the traditional volume datastores appear in PowerStore. You must utilize vVols to see this kind of data within the PowerStore Manager web interface. Creating datastores also requires you to manually provision the volumes that will be used, map them to the ESXi hosts, then manually create the datastores from the vSphere Client and grant host access. Don’t get me wrong, this process is not unlike that of most other storage systems. However, knowing how easy it was with the Unity family makes me hope that Dell brings this functionality and simplicity that the Unity provided to the PowerStore solution.

Final Thoughts

The PowerStore is still on version 1 and I have only had a couple days to play around with it, but it still seems like a solid platform so far that I’m looking forward to implementing for a number of customers. It’s a significant step up for my lab from the VNXe 3200, and will be a significant step up for most of my customers as well. I’m eager to test out the performance and additional capabilities the PowerStore has to offer, but my first impression is definitely positive overall.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s