Dinesh Singh

Hitachi + VMware vSAN + Intel Optane NVMe = Turbocharged Hyperconverged Infrastructure

Blog Post created by Dinesh Singh Employee on Sep 26, 2018

Today, Hitachi Vantara announced a new hyperconverged system designed to provide high performance to businesses looking for faster access to data with a software-defined architecture. Hitachi partnered with VMware and Intel to co-engineer an all-NVMe hyperconverged Unified Compute Platform HC system, which leverages Intel® Optane™ SSD P4800X, next-gen system accelerator built on Intel® 3D XPoint™` technology.

 

IT teams have no choice but to resort to excessive resource deployment to meet application performance SLAs. Complaints from VDI users, frustrated with performance degradation are all too common. Today, all-flash HCI systems, based on SAS and SATA SSDs, helped reduce the performance issues, but may not totally eliminate them.

 

Growing data volumes and evolution in analytics techniques necessitated a radically different approach to manage data that bridges the gap between processor and storage. By eliminating the SAS controller-induced latency, a new storage access technology, NVMe protocol, is transforming the way we store, access and monetize our data. NVMe is a fast data access protocol that rides on top of the PCIe bus that helps unleash the potential of your business data and drives profitable business outcomes through faster insights from data. NVMe storage takes performance to an entire new level and allows IT teams to better harness data to derive insights for better business decision making.

 

NVMe offers a Performance Boost for Virtualized Environments–

       Picture1U.png

Hitachi UCP HC, with Intel® Optane™ SSD DC NVMe P4800X with NVMe drives in the caching tier and the Intel® SSD DC P4510 with NVMe in the capacity tier, delivers breakthrough performance enabling up to 3X improvement in IOPS and up to 4X improvement in storage latencies. Hitachi and Intel engineering teams collaborated to run joint tests over many months to determine and optimize the performance results for real-world scenarios. Teams performed tests on VMware’s HCIBench and HammerDB to replicate actual use case scenarios and observed superb performance enhancements over SATA all-flash systems. The following charts show the impressive test results under virtual server use case (HCI bench) –

 

                       

 

  1. 225% higher IOPS when multiple VMs write into persistent storage
  2. 3X higher IOPS for write-intensive workloads
  3. 70% reduction in disk latency for write intensive workloads.
  4. 225% higher throughput
  5. 4X higher IOPS/$

 

Hitachi UCP HC with Intel® Optane™ SSDs Doubles Performance for Relational Databases -

 

  

 

Hitachi’s all-NVMe UCP HC opens new possibilities for customers running revenue-impacting relational databases. NVMe drives dramatically reduce latency up to 84%, making it an ideal platform for performance-intensive workloads such as SQL, DB2 and other relational databases. Intel® Optane™ SSDs in the cache results in almost 100% increase in number of transactions, doubling the scale to cover more transactions to respond to revenue growth, for customers such as  retailers, hedge funds, credit analysis firms. It can analyze more data and help gain insight faster to respond to dynamic business conditions.

 

Hyperconverged infrastructure is a distributed, scale-out architecture, which means the overall performance is determined by all of the components in the vSphere clusters including CPUs, cache, persistent storage and network. Hitachi UCP HC brings it all together for you to minimize any performance bottleneck by integrating Intel’s sixth gen processors, 25GB port speed with NVMe cache and capacity tier to ensure faster data writes, retrievals and high throughput.

            

Recent advancements in storage technology has presented a clear use case for hybrid, all-flash and all-NVMe UCP HC systems. The image below illustrates the optimum use case for each solution.

 

               

 

Hitachi has taken a pragmatic approach to bring to the market, storage solutions based on all-NVMe architecture, prioritizing the acceleration of hyperconverged infrastructure instead of engineered VSP storage systems that offer specific hardware accelerated flash modules. Introduction of Optane-based NVMe on hyperconverged series allows customers to embrace a software-defined approach to drive performance on cloud-like just-in-time IT consumption model. Continuing the Hitachi tradition of providing choice, performance and unerring resiliency to our customers, we plan to offer NVMe in our all-flash VSP family (and thus our traditional converged infrastructure solutions) in 2019.  More information about Hitachi’s view of NVMe and our plans for adoption can be found in a recent blog by our CTO, Hu Yoshida, which can be found here.

 

You can see live demo at Hitachi booth #P311 at VMworld Barcelona in a few weeks. Feel the excitement with Hitachi Vantara !

 

Outcomes