Storage Predictive Analytics

Let’s start with understanding about storage predictive analytics. This term is combined  from three terms. Firstly, Storage which refers to data storage appliances and devices . Secondly, Predictive refers to predict  some metrics based upon the data availability. Thirdly, Analytics relates to analyzing the data stored with help of algorithms. To summarize Storage Predictive Analytics helps you to perform analytics onto your  storage systems.

There are several challenges faced by Storage vendors today. Firstly it is very difficult to figure out root-cause of failures. L1 and L2 support teams struggle a lot while  triaging root cause failure of storage appliances. Storage Administrators do not get a granular level of performance analysis of storage appliance. Triaging for Hardware and software failures. Analysis whether resources are working optimally or not. Interoperability issues and configuration issues are also part of major challenges faced by industries.  Less throughput and increased latency in accessing applications. Knocking doors of various vendors to get root cause analysis of failures. All these challenges causes overall result in experiencing downtime and data loss.

Storage Predictive Analysis helps to overcome these challenges in many ways . Firstly the self management, self healing , self optimization and auto supports. Data center administrator gets access to a single page application which provides granular detail regarding how storage arrays and storage server are performing.

This technology could be broadly divided into 3 parts– Extraction of data, Analyzing the data with help of customized algorithms and analysis on stats, which is a continuous process. Extraction of data could be with help of sensors which could be either logical sensors or physical sensors. Logical sensors are algorithms written which provide analysis on thousands of data stats; and physical sensors could give you information of temperature and fan speed in data centers. Finally storage administrators get access of Graphical User Interface which shows detailed analysis of their data centers.  

Most vendors shows policy and schedule  information regarding the data replication. But overall data protection analysis is missing which could be provided with predictive analysis . for instance, let’s say we can predict  total percentage of volumes protected with help of number of snapshots taken , which could be taken locally or replicated to some another geographical area.

Also recommendations are provided when to take snapshots and how to make your data protection policies more efficient. In short Predictive Analytics  can learn from historic data and can give you optimization recommendations .

When we talk about data protection, two important terms RPO and RTO come to mind. RPO stands for Recovery Point Objective and RTO stands for Recovery Time Objective. The aim to to ensure that RPO is zero at all times, so that there is no data loss. Also in ideal case  RTO should be zero so there is no downtime. And to achieve this you need to have smart replication policies considering minimum performance impact when replication or backups are in progress. Predictive Analysis can suggest you when we can trigger our replication policies.  

To summarize we can say Storage Predictive Analytics eliminates need of L1 and L2 support after selling storage box by automated root-cause of failure in an automated way and helps to give insights to customer about utilization of storage,capacity prediction and reports such as data protection report,capacity reports.