Data Center Advisors

What happens if you have a goal of creating a 1000PB scale community or service for this newly emerging cloudy world?  Is spinning disk really the right thing to consider for these extreme storage environments?  (A big thank you to Ken Wood for that term I’m going to steal, err adopt for this piece.)  Hitachi has been regularly getting requirements from two camps those who want to get rid of all tape and backup software in their enterprise and those who want to continue using it.

The reasons for removing tape from an enterprise really come from the continued lengthening of  backup jobs and the risks associated to keeping removable media.  For instance when a discovery order comes in companies are required to know where all of their data is (even if it is in a storage vault with water dripping on it) and if the policy says it is accessible then it had better be.  Because off site storage frequently gets used for removable media some times enterprises don’t know all of the copies of the data that they have and end up continuing to find information post discovery which basically makes them look bad in the eyes of the court.  Remember the amended Federal Rules of Civil Procedure simply state that companies have to have a records retention policy and adhere to it.  So that “soggy” tape is germane for discovery and if you cannot get the information off of it you could be in a world of hurt.  The knee jerk reaction is to just get rid of all of that tape including the ones that end up accidentally in dumpsters revealing sensitive customer information. However is this right?

Hitachi’s work with customers that need extreme storage environments, thanks again Ken, has uncovered a very interesting use case that I want to explore as the counter argument to merely getting that tape monkey out of the enterprise.  Specifically if you can imagine a company who produces map information like what is available Google Earth, and let’s say that a natural disaster like a flood occurs then there is a legitimate question to ask from relief NGOs and agencies: can said company help me see what the terrain and housing looked like prior to the disaster?  Well if the information is accessible via offline media like tape then it would be possible to bring that information back into the live system for review in a specialized domain by the relief organization.  (Mind you I’m not making this use case up, it is a real one.)  The rationale for a company using tape in this case is that they did the TCO calculation for keeping all of that data on spinning disk and found out that the energy consumption and floorspace required made their application too expensive to deploy; however, should we say to the NGOs, sorry we don’t use tape so we cannot keep information that long?

Similarly with customers who are looking at deploying an exabyte they too are concerned about how to pay for the associated costs related to cooling, power, floorspace, etc.  Basically they are saying that they can only afford environments which have 90% or more capacity on tape at the exabyte scale.  So I guess I’m seeking some collective input from the Internet to see what the direction is?  Mind you I’m not trying to lead anyone down a specific path, I honestly am asking the question.