How ASEAN firms can tap storage analytics to improve operations
With storage systems becoming more heterogeneous and siloed, some organisations in ASEAN are turning to storage analytics to reduce costs and address storage issues before they occur
Coupled with the data deluge and the fact that storage footprints often comprise different storage systems and management tools, underlying issues in the storage infrastructure can often remain undetected – until an outage occurs.
“As storage systems become more heterogeneous and siloed – by application, location or tier – the more diverse the management tools and processes become,” says Wayne Pauley, senior analyst at Enterprise Strategy Group.
“This can often force storage administrators to opt for manual control via the CLI [command line interface] and dilutes their ability to find and solve problems in a timely manner.”
The solution appears to be in storage analytics, a set of technologies that could be powered by artificial intelligence (AI) and machine learning (ML) capabilities to help storage administrators predict and address storage issues before they occur.
“Storage analytics has greatly simplified the process to periodically gather required telemetry data from various infrastructure resources – such as server, network and storage – and generate a performance trends report for various workloads running in an enterprise,” says Ang Kiat Wee, technical expert for data management and protection at Hitachi Vantara Asia-Pacific.
Previous approaches of monitoring alert parameters have contributed to a phenomenon where users see a sea of red alerts on monitoring dashboards or experience thousands of alerts attributed to an “IT event storm”. This creates confusion for storage administrators, who have to figure out which alerts to focus on.
“With storage analytics, not only can we proactively identify performance problems early or situations whereby an application will soon be running out of capacity using predictive analytics, we can also define actions to quickly remedy the problem through automation,” says Ang.
“This can further eliminate problems before they affect users, dramatically reduce troubleshooting times and help customer achieve reduced mean time to repair [MTTR] objectives for their IT operations.”
But despite its promises, storage analytics services – at least the cloud-based ones – are not widely adopted in ASEAN today.
“Most AI/ML-powered cloud-based storage analytics services are very new to the market and often require a client to move to a different cloud storage service,” says Tim Sheedy, principal advisor at Ecosystm, a Singapore-based technology consultancy.
“For example, AWS’s S3 intelligent tiering is a different storage class from the other S3 storage tiers which have no ‘intelligence’ beyond the standard analytics,” he adds.
“If a company clings onto a five-year capital expenditure purchase cycle and slow provisioning and change control procedures, it will find it difficult to realise the full value of storage analytics”
John Martin, NetApp
For now, the biggest users of storage analytics are major global organisations, according to John Martin, director of strategy and technology at NetApp Asia-Pacific’s CTO office. “We hope to see more local small and medium-sized enterprises follow suit soon,” he says.
In ASEAN, organisations in industries that are already using storage analytics, such as telecoms and banking, have started to integrate the technology into core processes, says Ted Aravinthan, director of modern datacentre at Dell EMC South Asia.
“Most of our customers interlink the storage analytics processes to support contracts – enabling them to respond in a much quicker manner when resolving any issues that may surface in their day-to-day operations,” he adds.
Meanwhile, there are plenty of opportunities for ASEAN organisations to further tap storage analytics to get the most out of their IT infrastructure, says NetApp’s Martin, noting that it is currently rare for a company to have detailed historical knowledge of its storage performance, reliability or capacity consumption.
“Most enterprises are still practising traditional purchasing behaviour – they buy as much as they can afford every three to five years according to their budget cycle, hoping that it will be enough. If there is a performance problem or they run out of capacity, they will typically start doing a bunch of reactive troubleshooting,” he says.
Cost savings
Since storage analytics can record historical infrastructure capacity growth trends and make future growth predictions, one of its main benefits is its ability to automatically move storage to different tiers based on past, current and future access requirements.
Sheedy notes that this could mean significant savings on storage costs or improvements in processing time – since the data does not have to be moved back from less costly tiers of storage.
For example, an action of a customer, such as kicking off a warranty claim, could trigger a series of processes to get data back from cold storage to live storage to process the claim.
“A normal process would have been for the customer service agent to trigger this retrieval of data, which can take hours or days,” says Sheedy.
NetApp’s Martin says the ability to implement effective tiering also helps IT teams to identify which business entities can benefit from higher performance and more expensive storage – or which are consuming more than their fair share of limited performance capabilities.
This lets organisations avoid over-provisioning and implement just-in-time purchasing, driving down capital costs associated with infrastructure and helping IT teams identify where and when additional storage resources must be purchased or allocated.
Pitfalls to avoid
Martin, however, warns that storage analytics tools will sometimes give recommendations that are challenging for the IT storage team to execute.
For instance, capacity is generally easy to add at reasonable costs with little disruption, but enhancing performance often requires significant work.
This is partly why all-flash arrays became popular. Martin notes that by providing extreme levels of performance (even if it is more than what a typical enterprise will ever need), solid-state storage reduces the need for IT teams to stop thinking about managing or measuring performance until a problem arises.
“Having said that, to make it affordable, a fair number of assumptions about the effectiveness of deduplication and other storage efficiencies such as thin provisioning have to be made, causing customers to begin keeping a closer eye on capacity consumption,” he says.
“When we work with clients who are looking at building a service-oriented infrastructure or private cloud, where agile provisioning and cost transparency are important, one of the first things we do is to use our monitoring tools such as OnCommand Insight to analyse their workloads and help them create a service catalogue that is aligned with their business needs.
“During that analysis, we often find areas where the customer can make significant cost savings, improve efficiencies and improve service levels. Overall, the best return on effort for storage analytics comes when it is aligned to support higher-level IT workflows and asset management.”
Noting that enterprises today are often held back by siloed analytics systems, including data warehouse, data lake, streaming analytics and AI clusters, Pure Storage recommends building a robust, efficient and scalable data-centric architecture.
“Modern data requires a new class of scale-out storage – a data hub built for unstructured workloads with cloud service-like agility and simplicity,” says Chua Hock Leng, managing director of Pure Storage Singapore. “This is why Pure Storage introduced our data hub architecture, which integrates the most important features of these four silos and unifies them into a single platform.”
Security concerns
Some infrastructure analytics platforms, such as HPE Infosight, collect and analyse sensor data points every second from a global customer base. Typically based in the cloud, although on-premise versions are also available, these platforms continuously learn from the installed base, making infrastructure – including storage systems – smarter and more reliable.
“Most of our customers interlink the storage analytics processes to support contracts – enabling them to respond in a much quicker manner when resolving any issues that may surface in their day-to-day operations”
Ted Aravinthan, Dell EMC
However, Paul Haverfield, hybrid IT CTO and pre-sales lead at HPE in Asia-Pacific, notes that ASEAN enterprises still hesitate to allow outbound connectivity from datacentre infrastructure to cloud-based monitoring (call-home) services out of security concerns. They are also likely to dismiss any kind of remote support solution without fully evaluating the risks or benefits.
A small amount of time invested in understanding the security models and risks can save huge amounts of time and deliver significant benefits, he adds.
“The HPE customer base using the full breadth of our storage analytics capabilities has consistently reported extremely positive feedback to us. We have cases in Singapore where HPE InfoSight monitoring has detected failures with air conditioning systems where the customers own monitoring had failed to detect.
“Most commonly, the feedback from storage-centric IT admin folk is that HPE InfoSight has allowed them to easily highlight problems with overloaded servers, insufficient memory and poorly behaving applications, as compared to traditional methods of performance troubleshooting.”
Leveraging insights
For enterprises to get the most out of storage analytics, it is important to integrate insights generated from the analytics into daily operations instead of using insights on an ad-hoc basis. This will go a long way in supporting automation efforts to address challenges, says Dell EMC’s Aravinthan.
Aravinthan also encourages enterprises to look at different ways to take advantage of the insights generated from storage analytics previously unavailable to them.
This may require establishing new processes to fully maximise the value of these insights for operations, such as adjusting the logistical details of back-end jobs to suit a company’s requirements and resources – or even eliminating some back-end jobs that may be redundant.
Martin says enterprises should also realise that for storage analytics tools to generate additional value for the business, they need to be supported by the right people and processes. Employees, for one, have to be empowered to act on the insights.
“If a company clings onto a five-year capital expenditure purchase cycle and slow provisioning and change control procedures, it will find it difficult to realise the full value of storage analytics,” adds Martin.
Even as Aussie firms such as Bauer Media and Redbubble are lapping up object storage, others are facing visibility and analytics issues with the technology.
Airtasker is using Amazon Web Services to ramp up its storage capabilities to meet the needs of an expanding user base.
