HPE Nimble Storage dHCI radically simplifies infrastructure for applications by reimagining HCI without limitations.
Hands-on with HPE Nimble Storage dHCI
One of the more interesting technologies currently being deployed in the datacenter: disaggregated hyperconverged infrastructure (dHCI). In particular, we looked at HPE’s implementation of dHCI as they are a leader in this technology. dHCI is similar to hyperconverged infrastructure (HCI) in the sense that it allows storage, compute, and networking to all be managed from a single management plane (in HPE’s case, from vCenter Server); however, unlike HCI, dHCI does not need to deploy storage in lockstep with compute. To get a better understanding of storage in a dHCI environment and how HPE’s dHCI solution has automated and simplified the process of setting up and managing dHCI, we deployed it in an environment that had existing vCenter Servers. We felt that this would replicate the experiences users would have when doing an initial dHCI deployment. Our initial dHCI cluster will be comprised of two compute nodes connected to an HPE Nimble Storage array and managed using vSphere with the HPE dHCI plugin. To connect all the systems, we used an HPE FF570 32XGT. This switch has 32 10Gb Base-T, eight 10Gb SFP+, and two 40 Gb QSFP+ ports. dHCI vendors have consciously uncoupled the storage from the compute to provide datacenters the freedom to grow their deployments holistically, thereby preventing the stranded resource problem that is prevalent with HCI deployments. This imbalance with HCI deployments is due to very few applications growing compute needs at the same velocity as storage.
How We Tested
To get a better understanding of storage in a dHCI environment and how HPE’s dHCI solution has automated and simplified the process of setting up and managing dHCI, we deployed it in an environment that had existing vCenter Servers. We felt that this would replicate the experiences users would have when doing an initial dHCI deployment. Our initial dHCI cluster will be comprised of two compute nodes connected to an HPE Nimble Storage array and managed using vSphere with the HPE dHCI plugin.
To connect all the systems, we used an HPE FF570 32XGT. This switch has 32 10Gb Base-T, eight 10Gb SFP+, and two 40 Gb QSFP+ ports.
Compute nodes:
- HPE DL360 Gen10 servers
- Dual Intel Xeon 6130 procs, 128GB of RAM
- Redundant drives for the OS
- VMware ESXi 6.7u1 & Nimble toolkit pre-installed
Storage:
- HPE Nimble AF20Q array
- 12 960GB SSD drives
- 5.8 TiB of usable storage
- 10Gb ports
- 2 used as iSCSI targets
- 2 used for management
Innovation
HPE Nimble dHCI Configuration Checker
HPE Infosight
Automation
To be honest, we were somewhat surprised at the depth of automation and integration HPE has put into this solution regardless if it is an existing deployment or a new deployment. It took HPE less time to set up an entire dHCI cluster, including setting up a vCenter Server, than it would have taken us to set up, configure, and integrate a vCenter Server with a SAN storage array.
Benefit
The big draw of dHCI is that day-to-day operations – such as monitoring, maintaining, and adding storage and compute, as well as checking to see if the cluster’s configuration is correct – can be carried out from a central management plane; in HPE’s case, this is accomplished with the dHCI plugin on vCenter.
Conclusion
dHCI has the potential to be a game-changer in the datacenter as it blends the simplicity of HCI management with the flexibility of deploying storage and compute independently of each other as is done with a traditional datacenter. But dHCI can only be a game-changer if it is implemented correctly (i.e., not simply slapping together a storage array with servers). Instead, it involves engineering a hardware and software solution that can be fully managed, from deployment to day-to-day operations, in a holistic manner from a central management pane – and it looks like HPE has accomplished just that with HPE Nimble Storage dHCI.
The ease and simplicity of deploying the initial dHCI cluster, adding storage and compute, and upgrading the entire system were truly impressive. Even more, HPE has bundled in their vast knowledge of vVols and seamlessly integrated them into this solution. We also see HPE InfoSight as an invaluable tool to ensure that the system is, and will continue to be, unimpaired and encumbered by support issues. HPE Infosight makes this possible with their AI’s enhanced predictive support and preemptive recommendations, which allows for proactive – not reactive – system management. In short, HPE has done dHCI right.