This blog originally appeared at Blue Hill Research.

Amazon held its second annual AWS re:Invent conference in mid-November this year. It used the event to announce a number of new services and feature enhancements. Potentially the most compelling of which, were those concerned with advancements in Amazon’s Redshift service.

Redshift is Amazon’s data warehouse-as-a-service offering that allows customers to move their data warehouse to the cloud at a fraction of the cost of traditional options. Amazon advertises that it’s cheapest package equates to a cost of roughly $1,000 per terabyte per year, roughly 10% of what Amazon claims to be the market rate. (Table 1).

Redshift Pricing Scale

At its core Redshift is a storage solution. However, Amazon demonstrates an emphasis on cloud-based analytics. Amazon is partnering with business intelligence (BI) providers including ActianActuateBirstJaspersoftLogi Analytics and Tableau to make them compatible to run analytics in the Redshift cloud. Cloud-based analytics is an emerging trend, which Blue Hill has commented on before. A number of startups have begun to converge on the space as well, including Treasure Data and BIME, which specifically positions as cloud-based Big Data provider. Redshift offers some validation for market demand for cloud-based analytics, although Amazon appears focused on playing a support role and remaining “vendor agnostic” regarding analytics.

Amazon’s primary focus is on storage. The company is betting that enterprises are hungry for an opportunity to outsource the responsibility of data warehousing and slash the costs of storage. To that end, Amazon has designed Redshift to meet a mass consumption model. That is to say: Redshift is a “one size fits most” solution intended to fit the largest common denominators of organizational needs. This means that Redshift should meet most companies’ needs. However, what it gains in broad compatibility and scalability, it gives up in highly tailored needs of specific industry niches. As a result, Redshift is poised for rapid adoption, but will be kept from penetrating all corners of the market.

Nonetheless, Redshift is well positioned to meet many organizations’ data warehouse needs. The scalable cost options are compelling. However, they should not be the only factor organizations consider. Blue Hill recommends that organizations consider five criteria in their evaluation of Redshift:

Base Costs: The per-terabyte savings of Redshift are compelling and potentially disruptive to the data warehousing market. However, organizations considering taking advantage of the analytics plug-ins will find that Redshift also charges an access fee to use the partner BI tools in their cloud. Doug Henschen of InformationWeek estimates this is around $0.16 to $1.80 per hour depending on computing power needed, even if the licensing of the software might otherwise be free. These potential costs should also be included in an organization’s evaluation.

Migration Costs: To take advantage of Redshift, organizations will need to migrate their data to the cloud. Migration is generally a painful and expensive process that is best avoided. Redshift imposes a one-time migration cost. Amazon’s promise is that after initial migration, scaling up and down will be an easily managed process where decisions are independent of integrating costly new infrastructure. While the first move to Redshift may likely be a massive undertaking, Redshift will help organizations to limit future migrations by eliminating capacity limitations.

Loss of Control: Redshift frees companies from architecture and archive management responsibility. While this means major cost savings, it also means giving up a level of control that not all organizations are prepared to accept. In particularly, organizations with highly regulated or sensitive data management requirements may find that customization is required to make Redshift meet their needs.

Security: For many organizations, cloud solutions present security concerns. In part, this relates to perceived vulnerability of cloud data that aren’t always accurate. For most organizations, Amazon’s security protocols are more than sufficient. The deeper question is: to what extent does your organization need to be in charge of data security? Amazon possesses the incentives and resources to monitor and adapt to changing standards in data storage and security that exceeds the capacity of most organizations. However, those that face privacy demands to control data access, such as HIPAA-regulated healthcare organizations, may not find that Redshift is able to meet their needs.

Support: Migration to a mass market solution like Redshift often requires a sacrifice of personalized customer support. For Redshift to work, it must be available to a wide variety of data environments on a basically templatized basis. For many organizations this may mean adapting internal processes to the tool, rather than customizing the tool to their processes. An in-house database administration team has an advantage due to their awareness of business context. Support with context allows for response times are compressed and otherwise overlooked road blocks can be addressed. These tradeoffs must be assessed as well. Depending on an organization’s needs, Amazon’s standard resources may be sufficient or they may want to inquire about dedicated support options.

Individually, none of these factors alone should be the decision-maker for organizations considering Redshift. Together, they should constitute a starting point for assessment. The next step is to assess the total cost of ownership of current warehouse capabilities. This includes not just personnel, hardware, and infrastructure costs, but also often overlooked issues like: electricity and disaster recovery contingencies. Much of these can be avoided with a hosted service such as Redshift. By comparing the balance of these savings with the costs related to cloud storage, organizations will be able to make the right determination for themselves.