As the industry inevitably marches towards a widening set of cloud services, the layers of the IT onion separate and render themselves “as a Service.” Ironically perhaps, the industry started at the highest layer, applications or SaaS. But PaaS is gaining momentum, evidenced by nearly all the most well-known public clouds exhibiting a variety of PaaS and BI/analytics individual service slices, such as:
DBaaS = Database-as-a-Service
INTaaS = Integration-as-a-Service
DEVaaS = Development-as-a-Service
HaaS = Hadoop-as-a-Service, and its close cousin
BDaaS = Big Data-as-a-Service
Will the “aaSes” ever stop? Not yet, for two WaaS, Warehouse-as-a-Service, offerings were unveiled yesterday, November 28, 2012, both specifically running on Amazon Web Services (AWS) and both unveiled at the AWS re:Invent conference.
AWS, itself, was one of the two vendors introducing WaaS, with Redshift. Light on architectural detail and heavy on media blitz, initially Redshift will be available in two node types: One for dipping your toes with less than 10TB of direct attached storage, the other with more substantial storage capacity that approaches the type of node you would expect for near-enterprise data warehouses. AWS claims you will be able to scale your nodes up to 1.6PB of compressed user data with “a few clicks of the AWS Management Console or a simple API call.” AWS also states that there is no charge for data transfer to/from Redshift, but that only applies to data to/from AWS regional sources, meaning you will pay for data loads originating from on-premises, appliances and other clouds. Jaspersoft and MicroStrategy announced certification for RedShift.
Despite the hype surrounding the Redshift announcement, I find at this early stage that the just-emerged-from-stealth BitYota offers a more fascinating WaaS. BitYota, barely a year old, but already supported by $12m from a longish list of VCs including major big data backer Andreessen Horowitz, offers a fully SQL-92 compliant data warehouse, but there is more to the story.
One of the first questions to ask any vendor with an “aaS” offering is, “Are you really multi-tenant, or are you just a hosted service with a browser front-end?” It may not always matter to the buyer, but ESG believes that in WaaS there is an opportunity to differentiate by truly taking advantage of the native cloud infrastructure. BitYota has latched onto that key concept, perhaps setting them apart as a first pure-play mover in WaaS. Cloud services, of any letter combination, require a flexible, perhaps parallel, certainly elastic infrastructure. What if you designed and built a data warehouse expecting that type of cloud infrastructure, rather than cloud-adapting a data warehouse originally designed for older architectures?
What you would produce is a true cloud MPP-style warehouse, with a purpose-designed SaaS UI, that includes fundamental one-click data integration. BitYota understands that they are not likely to displace, for example, many or any Teradata or Oracle data warehouses, or displace Informatica as a cloud integration alternative. However, BitYota can add value as an adjunct warehouse to deal with semi-structured data, with an underlying massively parallel engine spun-up with a click. The SaaS UI enables users to define and set data policy and visibility rules, and to load data through a built-in scheduler using an automated engine to detect schema, frequency of change and other pertinent data load characteristics. BitYota uses the in vogue RESTful web APIs to programmatically talk with the outside world.
I see three use cases where BitYota may help organizations with their warehouse and big data efforts, and have a suggestion regarding what BitYota needs to concentrate on next:
As already mentioned, (1) BitYota may serve as an adjunct to existing warehouse implementations looking to expand along the data variety axis, and/or (2) looking to escape the on-going drudgery of infrastructure management for on-premises data warehouses, either in terms of space or speed. But also, (3) for those involved in their first big data projects, and feeling confounded by Hadoop-as-a-Service, BitYota offers a means to offload complexity and add cloud scale, acting as an analytics repository between on-premises data and a big data-as-a-service solution.
The next trick for BitYota involves alliances.There are about 125 vendors in ESG’s list of legitimate vendors directly involved in big data. Despite the desire of every one of those vendors to dominate share of wallet, such dominance is a rare phenomena. Generally, the big data vendors that play well with others, particularly earlier stage vendors, augment their chances to at least survive if not thrive. While RESTful APIs and SQL-92 are excellent architectural choices, they are not likely to singularly entice many buyers for BitYota. Even in the cloud there are technology and go-to-market partnerships, and BitYota will need to pursue such partnerships immediately for the market will quickly get croweded. There are already other hybrid-WaaS offerings from the likes of Teradata who already offers its Active Data Warehouse in private cloud mode and Teradata Express on AWS. And Oracle is targeting its purpose-built Oracle Exadata X3 warehouse-oriented appliance through its own cloud.
BitYota’s unique technology approach notwithstanding, as BitYota emerges from stealth it must rapidly reorient itself toward ecosystem and channel, a typical shift facing most early stage companies, but in the big data market, time is particularly precious. Let us assume for the sake of analysis that BitYota’s WaaS IP is more sophisticated than AWS Redshift, but the AWS Redshift certification from Jaspersoft and MicroStrategy and related media will draw more eyes. BitYota may never have the MarCom or alliance management budget of an AWS, Oracle, or Teradata, but BitYota still must engage in the air war.