Innovating Data Warehousing With Cloud Technologies

What impact is cloud technology set to have on data warehousing?


Data warehouse analytics have not always been seen as compatible with cloud functionality. The nature of analytics means it borders on being a bespoke field, with each company's analytic practice differing if even only slightly - traits at odds with the no-assembly-required nature of the cloud. There is also the public nature of the cloud, with some organizations reluctant to store their data there as a result, believing that it is better secured if housed internally. On top of this, cloud technology is not really configured for typical data workloads.

A number of innovators in the industry are, however, challenging these ideas and developing new ways to introduce data warehouses onto the cloud in a way that navigates such issues and takes advantage of its numerous benefits.

The traditional data warehouse is not built for agility. It has a fixed capacity, and offers a set amount of space to store data, for which a firm pays far in advance. Conversely, the benefits of the cloud lie in its inherent flexibility. The cloud has an elastic capacity, and therefore a completely different nature to the traditional data warehouse. Reconciling the two requires looking at data warehouses in a whole new way.

Firms such as Snowflake Computing are among those to begin introducing systems which exploit the cloud’s elasticity and embrace its capabilities. Companies are then able to save heavily with the cloud’s pay-for-use model, as they no longer have to pay for space in advance. They have also cut deployment time down to nearly nothing in keeping with the cloud's functionality, and made support easier to access for any scale of users and workloads at any time.

In Snowflake’s cloud-based data warehouses, the sizing, provisioning, and scaling are all automated. They have moved away from the MPP models used by Hadoop and others, in which both paradigms combine scalable storage with scalable parallel processing. Instead, computing and storage are split entirely, removing the need to size and balance a system before it is deployed. By building systems specifically for the cloud, as increasing numbers of firms will do, it makes it more elastically scalable as normal MPP technology - and at a fraction of a cost and in far less time than with a traditional data warehouse.


comments powered byDisqus

Read next:

Social TV: Cross-Channel Insights on the ShareThis Platform