Will cloud data programs become the future of DataOps?
Within enterprises, legacy platforms are becoming marginalized as modern data-driven platforms become the preferred choice for data teams. Artificial intelligence and machine learning are powerful solutions for big data that are encouraging enterprises to accelerate their digital transformation towards the cloud. It should be no surprise then that, according to Cisco, four percent of workloads will be hosted in cloud data centers by 2021. Despite this, there is some reluctance amongst organizations to build on their data programs with these solutions. Regardless, ignoring the cloud is a critical oversight for enterprises looking to meaningfully analyze the vast quantities of structured, semi-structured or unstructured data within their networks. More concerning, however, is the critical insights that will be overlooked or missed entirely by enterprises relying on legacy software.
The cloud is starting to clearly denote itself as the de facto choice for investment in big data. Canalys estimates that cloud investment will surpass $143 billion by 2020. While Fortune 500 companies have historically been reluctant to dip their toe in the digital transformation pool, there has been a radical shift in attitude in recent years. More than a corporate buzzword, the term 'digital transformation' now carries with it the promise of large ROIs and even larger data pipelines. This has lead to a culture where having large-scale, full production workloads is a tangible reality and not merely a distant goal.
What can the cloud offer big data?
The workloads processed by modern data applications demand elastic scaling. What this means is that storage and compute needs are often changing independently from one another. For such workloads, the elasticity of the cloud is a distinct advantage. Cloud environments can ensure that data pipelines perform optimally regardless of these rapidly shifting requirements. For next generation data applications, this is no longer a niche requirement but a necessity. For instance, for social media or dating sites where there is peak traffic at certain times and minimal traffic in others, elasticity is essential. In a cloud environment, additional processing power can rapidly be deployed and rolled back where it is needed. Whether these spikes are on a daily, weekly or monthly basis -- the agility of the cloud cannot be matched by on-premises environments.
The challenge of "bursty data"
Unpredictable changes in data volume, velocity and variety present a unique challenge to data processing and analytics tools. This is perhaps most clearly demonstrated by e-commerce retailers who experience unprecedented spikes in traffic during major shopping events like Black Friday and Cyber Monday. During normal operations, e-commerce retailers will use data processing and analytics tools to generate real-time targeted insights for customers. During these events, the on-premises infrastructure used to facilitate these activities will be strained beyond capacity. To maintain operations, the retailer would typically have to deploy more physical servers to manage this spike,which would not see see use the other 364 days of the year. In a cloud environment however companies are able to increase scale on demand to match traffic surges. Essentially, a cloud data program is cheaper and requires less manual configuration.
Hybrid really is the best of both worlds
While the advantages of the cloud are clear, there are some instances where an on-premise environment may be more suitable. This is especially true of organizations with sensitive, personal information on their servers. For those concerned about the integrity of cloud-based security, the hybrid cloud model allows them to enjoy the best of both worlds. A hybrid approach combines the elasticity and scalability of the cloud while keeping their sensitive workloads on premises. This method is already gaining traction amongst enterprises with 56 percent of US IT decision makers using a multi-cloud or hybrid cloud strategy according to a recent survey.
For enterprises interested in longer-term data growth, flexibility and cost-saving, investment in cloud infrastructure should be their first consideration. They also need to make important decisions on whether their cloud-strategy will support their long term goals and if they could benefit from a hybrid cloud set-up. Regardless of what approach is taken, the cloud will play an integral and central role in forward-thinking enterprises.
Kunal Agarwal is CEO and Co-founder of Unravel Data. His past experience includes tours of duty leading sales and implementation of Oracle products at several Fortune 100 organizations and helping Sun Microsystems run its Grid Computing Engine. Started his entrepreneurial journey by co-founding Yuuze.com, a pioneer in e-commerce personalization.