By Jasmine Desai
While customer expectations from the storage market keep on increasing, storage technology itself is taking customers on a journey with its fast-paced innovations. It is a chain reaction set in motion wherein, an area like DR immediately affects data backup and in turn storage. As data growth escalates, so does storage investments with it. According to Gartner, the IT spending on data center systems in 2014 will US $1,443 billion. One of the most talked innovation this year in storage has been flash and SDS.
The flash and SDS combo
Flash is definitely the future of storage. Richard Fichera of Forrester Inc. mentions in a blog post that the introduction of flash in main memory DIMM slots, has the potential to completely disrupt how flash is used in storage tiers, and can potentially break the current storage tiering model, initially physically, with the potential to ripple through memory architectures.
There has been a lot of concern around potential architectural challenges when it comes to usage of flash. However, abating those concerns Akhil Kamat, Volume & Storage Leader, STG, IBM, India & South Asia says, “Flash does not have architectural challenges, because the way enterprises have segregated the data, is in tier 1 and 2. SDS gives the entire intelligent layer and flash gives the performance layer. Customers today know what they want to do with flash in their environment.”
Flash storage, with its fast accessibility will begin to replace tier-one storage in many data centers. These data centers will also look for new ways to maximize storage utilization on solid state storage to reduce cost of implementing flash storage. According to Santhosh D’Souza, Director – Systems Engineering, NetApp Marketing & Services Pvt. Ltd India, “We have seen a significant increase in solid state/flash storage adoption in recent years, to the point that flash storage will be tried and implemented at nearly every Fortune 1000 enterprise within the year.”
Also, the flash/SDD market still has a healthy number of start up organizations that are making strong in-roads into the data center. However, they are rapidly being consumed by larger traditional storage companies.
Thus, if customers have not adopted it, they are considering it very seriously. SSDs may be placed within a server, as a dedicated storage device on a shared network or most commonly, within a shared disk array.
On the other hand, Software Defined Storage (SDS) is an area of considerable interest to Indian enterprises, as data growth continues to outstrip all efforts to manage existing storage environments, and as organizations increasingly become data-centric, in their pursuit of top-line and bottom-line growth. Placement of SSD technologies within a storage infrastructure offers a variety of advantages and challenges depending on their proximity to the processor. According to Jim Simon, Senior Director of Marketing Quantum Asia-Pacific, “SDS has higher performance and higher reliability. Flash storage is becoming more affordable and therefore its adoption rate is increasing. Organizations are using it as a separate tier of storage by itself, as a primary storage and they use disk array for secondary storage and tape for tertiary storage.”
Also, the SDS software can be executed on any type of server hardware and does not have to be installed on dedicated hardware. Over-provisioning is a problem born out of a variety of constraints trying to deliver high levels of performance on a small data set for instance. And SDS will be the culmination of these phases kicking in to abstract and automate the virtualized data storage platform.
The deal around DR-a- a-service
The Forrester Wave: Disaster-Recovery-As AService Providers, Q1 2014 mentions that during the past several years, a slew of DRaaS offerings have hit the market with the promise of faster recovery in the cloud at the same or lower price points, and more flexible contract terms compared to traditional recovery methods. While this may sound too good to be true, many who have taken the plunge report that these claims are not as far-fetched as they seem at first glance. During the past few years, enterprise DRaaS adoption has grown steadily to 19% today, with another 22% planning to adopt.
According to Prashant Gupta, Head of Solutions, Verizon Enterprise Solutions, “There is certainly a market need for DR-as-a-Service. No one wants to invest in a capacity that is not utilized. Security is a key concern here in terms of access to the data, who has access to it, also where is the data going in terms of geography and what are the laws of the land the data is going to reside in.”
While providers can mitigate some of the security concerns through better identity management; data eradication; role-based access; and compliance with standards such as SSAE 16 SOC 2, ITAR and PCI, the real onus of ensuring data privacy still lies with the customers. DR as a Service also helps save organizations a lot of operational costs.
Citing an example, Kamat of IBM says, “Everyone needs to be prepared for DR in an organization right from the process piece of it. There was a DR drill done in an organization and a fake rumor of earthquake was spread. People who moved out of the building first, were the ones who were supposed to take care of DR. An organization can have the most sophisticated technology in place, but without a proper process in place, it does not amount to much.”
The provider who gives the flexibility to choose geography, will have much more value in this area. Says Simon of Quantum, “It is mostly relevant for SMBs. They have had to buy backup systems traditionally that would not only fit their current needs but future needs. DR-as-a-Service becomes OPEX as one is paying on a monthly basis.”
With the recent spate of natural disasters, DR is becoming more and more a major focus for I&O professionals. Anand Ghatnekar, Senior Director, BRS, EMC India Center of Excellence says, “High-availability is the one and only one thing that I/O professionals need to remember while preparing DR from natural disaster perspective. The IT infrastructure environment in today’s data center is way too complex and high availability is a system design approach to ensure required measures are taken to avoid SPOF’s (Single Point of Failures).”
I/O professionals play a key role in eliminating the SPOFs by properly architecting the storage requirements in a data center. They can be eliminated using a combination of ‘redundancy’, ‘failover’ and ‘fallback’ mechanisms. This generally entails deploying high-availability storage solutions that utilize technologies such as RAID, multiple/redundant storage controllers, remote mirroring, multiple I/O paths, and clustered storage systems.
According to Nitin Mishra, Senior Vice President – Product Management, Netmagic, “When it comes to natural disasters and preparing the DR, location becomes very important. One has to ensure that the data is available and up-to-date. Location and operational readiness including DR drills and monitoring it on continuous basis is the key.”
Chandrasekaran of Gartner says, “Thus far, Recovery-as-a-Service (RaaS) has been adopted far more broadly by small and midsize businesses (SMBs) than by larger enterprises. A key reason for this is that large enterprises are more likely to have established recovery facilities, as well as the in-house expertise and mature processes needed for sustainable recovery assurance.”
He emphasizes that organizations should plan for disaster scenarios based on the likelihood of that scenario occurring. Scenarios based on criteria such as notification time (e.g., “a tornado warning is in effect starting tomorrow at noon”), type of disaster and potential business impact should be established only if there are material differences between the response and recovery procedures for that type of event. Unlikely (low-probability) events should be identified; however, rather than developing full-scale plans. One constant with good DR plans is that they are well-organized, easily navigated and easy-to-use.
However, organizations typically fail to update their plans frequently enough to keep pace with the rate of personnel changes associated with the individuals who are assigned recovery responsibilities. One effective way is to avoid the use of individuals’ names and contact information in the recovery document; and instead, use roles and job titles.
Data backup and recovery mash
The DR question inevitably leads us to look at data backup and recovery. This has been an ever daunting question that today’s storage and backup administrators are facing as mobile devices continue to proliferate data formation which is mostly in unstructured form. Unstructured data constitutes a large percentage — 75 to 80% of the overall data that we use on a day to day basis, while structured data constitutes the remaining 15 to 20%. Most of this unstructured data is in the form of emails, raw text, metadata, health records, file services (audio, video, images) etc. The volume of unstructured data has posed a huge challenge from a backup and recovery perspective. A good way according to Gupta of Verizon is, “To reduce the recovery time, firstly there needs to be categorization of data. Organizations should know what they are trying to protect or replicate.”
According to Kamat of IBM , “Unstructured data is huge. All of it cannot be stored on primary storage. Organizations should ask themselves what is the RTO (Recovery Time Objective) and RPO (Recovery Point Objective) they can afford.”
They need to keep in mind that no data protection technology can yield better recovery time if the recovery objectives are not well thought of. Mentions Simon of Quantum, “Even with data de-duplication, organizations tend to back up the same data. They also do incremental backup. One can opt for object storage,. The overhead associated with redundancy here is much lower. One does not need 100% redundancy here, but around 40%.”
There are many solutions that are available today from various backup vendors that can assist the back up of unstructured data. Mentions Ghatnekar of EMC India Center of Excellence, “ILM (Information Life cycle Management) is key to addressing this specific problem. There are many backup management schemes that are used in ILM such as de-duplication, retention and replication technologies that can assist with unstructured data backup depending on the RTO.”
These technologies assist in resolving the cost of storage and recovery time, but it really comes down to archival when companies need to decide what data they need to keep and where do they store it.
According to Chandrasekaran of Gartner, “The problem of the ‘shrinking backup window’ continues to be ever-present. Backup complements archive and archive complements backup theoretically. In reality, what Gartner continues to see is that these technologies are deployed in silos in most environments. IT leaders should use backup for operational recovery and deploy archiving for discovery and long-term record retention.”
There is a lot that goes into this planning considering the disaster recovery and business continuity aspects that an organization needs to have, and this may vary with organizations based on their business critical applications. Organizations need to plan upfront for their Recovery Time Objectives (RTO) and Recovery Point Objectives (RPO). These two parameters influence the kind of backup and recovery infrastructure that needs to be deployed.
jasmine.desai@expressindia.com