Hive Mind and Intelligent Infrastructure

By Hubert Yoshida posted 06-22-2020 19:52


Storage Newsletter just published an article by storage analyst Ken Clipperton comparing four storage approaches to offering Intelligent Infrastructure. He recommends that as companies evaluate enterprise storage offerings today, they should pay careful attention to these companies support infrastructure analytics and how they fit with their own needs, abilities, and culture. Multiple enterprise storage vendors are now offering what can reasonably be described as intelligent infrastructure. Their product roadmaps show how they are on the path to an automated enterprise which is the end goal for an intelligent infrastructure. He describes the four approaches by comparing four storage vendors: HPE Nimble Storage, DDN Tintri, Huawei, and Hitachi Vantara. Although he does not mention Dell Power Max, I would say that Dell’s approach is similar to what DDN Trintri.  


The four approaches that he describes are:


Cloud-based Analytics Integrated with Proactive Support
HPE Nimble Storage provides comprehensive cloud-based InfoSight predictive analytics capabilities that it tightly integrates with proactive support processes. HPE InfoSight has app-aware intelligence that considers real application behavior and makes recommendations driven by AI. InfoSight claims the complete automation of Level 1 and 2 support, with 86% of problems predicted and auto-resolved. This is primarily for maintenance and root cause analysis.


Analytics Integrated into the Storage OS
DDN Tintri VMstore focuses on consolidating huge numbers of virtualized workloads on a single storage infrastructure. It embeds analytics into its storage OS and uses those to manage the performance of every workload dynamically. Dell takes a similar approach through the use of machine learning to manage performance in the array.


Analytics Embedded in the Array via AI Chips
 focuses its analytics on enabling autonomous infrastructure. Its newest OceanStor arrays incorporate its own Ascend 310 AI chip. The chip is based on ML frameworks. It understands and actively analyzes the I/O rules of multiple application models, and dynamically adjusts to I/O activity. The firm claims this chip can improve the read cache hit ratio by 50% and shorten batch processing latency from 300μs to 150μs. The primary reason is for performance in the array.


Analytics Alongside the Array
Hitachi Vantara Ops Center approach is described by Ken as a multi-faceted operations and management software package. Ops Center runs on-premises in VMs or on a dedicated server. The goal is to maximize application performance. It applies AI to automate provisioning, data reduction, data migrations, and other data center workflows. Huawei supplements its AI chips with cloud-based analytics. It offers proactive support and predictive analytics via its eService Intelligent Cloud Management System, which relies on periodic collection and uploads of operations data, including alarms, configuration data, performance data, system logs, and disk information. This data enables intelligent planning, automated provisioning, monitoring, and end-to-end optimization.



Actually, Hitachi Vantara Ops Center’s goal is much more that maximizing application performance.


Hitachi Ops Center Analyzer improves IT operational efficiencies with AI operations. Manage, optimize, orchestrate, plan and protect Hitachi infrastructures as you simplify and automate IT processes. It can be described as an Alongside Array approach since it monitors the entire data path using telemetry. Hitachi realizes that you can’t manage everything from the array since the full data path is involved. IT infrastructure is made up of multiple interacting elements and an issue in any one area can cause ripples in the data center fabric that can reduce uptime, performance, and impact root cause analysis. Hitachi Ops Center Analyzer provides end-to-end insights and analysis on the data path, from VM or server through SAN networks to Hitachi storage assets. It delivers heterogeneous IT analytics to provide recommendations on managing system resources to optimize the full data path for business applications and leverage these operational insights to improve configuration, provisioning and data protection management practices. Hitachi Ops Center Analyzer provides predictive analytics capabilities to eliminate the guesswork that often accompanies new storage capacity purchases or new application resource deployments.  With forecasting using actual historical growth trends, you can improve capacity planning accuracy for better resource utilization and help reduce unnecessary storage purchases.



Hitachi Ops Center Automator improves IT operational efficiency by reducing the number of touches a human must make when implementing and managing infrastructure resources and data services. Designed to orchestrate the delivery and control of Hitachi and third-party infrastructure resources, Ops Center Automator minimizes the effort spent (fewer number of actions) to perform individual tasks by as much as 70%. The benefit extends beyond administrative savings. Infrastructure can be delivered 90% faster (wall clock) so that applications get the resource they need quicker, maintaining application uptime and ensuring the required performance. Customer satisfaction is increased, and SLA attainment is improved.

Another benefit is self-service. The storage administrator can create a service catalog of optimized services that are tailored to individual clients or business units. Each custom service limits the scope of the provisioning process, reducing the number of inputs that need to be defined which reduces the chance for human error. The introduction of NVMe and NMVe-oF 

is made simpler using Automator’s smart provisioning which leverages AI with Ops Center Analyzer to select the appropriate storage and fabric based on historical performance trends.


Many enterprises want to gain the benefits of automation and analytics, but their IT staff lack experience and their resources are limited. Hitachi Vantara addresses this need by offering a low cost “Automator Starter Pack.” This service helps the client to implement the software and then work through the process of automating two workflows with the customer across a period of approximately 90 days. The result is a quantified ROI and transfer of knowledge to enterprise IT staff that will enable them to automate additional data center workflows.


Hitachi Configuration Manager REST API - All the management capabilities provided by Hitachi Ops Center Analyzer and Hitachi Ops Center Automator can be accessed by the respective product GUIs or through standards based REST APIs for easy integration for reporting, IT service management or larger workflow automation with existing management application frameworks like vROPS, Lumada, Ansible, Kubernetes, ServiceNow, etc. 



Hitachi Vantara advises IT organization to buy into products with REST APIs that allow them to integrate with third party systems and tools, to create what Hitachi Vantara calls a “Hive Mind” based on collaborative AI. (Hive Mind refers to the collective thoughts, ideas, and opinions of a group of people (such as Internet users) regarded as functioning together as a single mind, for example - The tiny bees in a hive are more or less unaware of their colony, but their collective hive mind transcends their small bee minds.— Sven Birkerts and Kevin Kelly). Many AI based management tools or systems have limited awareness of how their actions could affect other systems. Customers should be aware of the potential for conflict and should set clear limits on which AI -based systems are allowed to control which resources.