Original URL: http://www.theregister.co.uk/2009/03/25/mimecast_fileserver_destruction/

Mimecast and file server destruction

Online mail services lead to filer destruction - maybe

By Chris Mellor

Posted in Storage, 25th March 2009 09:20 GMT

Analysis Mimecast is a UK-based supplier of unified email management services. Around the end of 2002 it started to build an appliance function that would unify the many daisy-chained email functions, (anti-spam, anti-virus, data leak prevention, signaturing) that were all implemented as separate boxes through which emails had to flow before ending up in peoples' inboxes. How it has done this and what it means leads logically through cloud file storage (CFS) to a deadly problem for filer suppliers.

The key thing chief technologist and co-founder, Neil Murray, decided was to put the actual storage in the cloud. Instead of using a client-side appliance or replicating customers' own email serving and storage functions in a Mimecast data centre, Murray decided to use a Google-like grid set-up. It uses Linux commodity servers, commodity filers, and Mimecast-developed software, with open source components, to provide an email repository that could scale massively and answer requests - where is Joe.Manager's subject X e-mail? - very, very quickly by parallelising the search across many gridded servers.

The result is a core platform with email-aware functions layered on top, which has proved attractive to many customers in the UK and abroad. Murray likens it to SAP, another core platform with business modules layered on it or plugged in, and says Mimecast will add file archiving this year. It will also add SharePoint support at some stage. Then Mimecast will have a single, multi-tenant-aware repository, that single instances file-level content and provides search, backup, restoration, archiving, compliance and general document management services in the cloud across the email, file and SharePoint domains.

Step back and consider Mimosa, which recently added the storage and archiving of SharePoint content onto its NearPoint archiving platform. This already covers Exchange email and unstructured Windows file archiving, with search, eDiscovery and compliance capabilities. It's another core platform with content-aware plug-ins providing for the capture of several data types and their storage in a central repository, followed by the provision of end user functions for search, eDiscovery, compliance, retention policies and so forth. Mimosa and Mimecast are parallel tracking in this super content management space, except that Mimecast is building its own system infrastructure in the cloud.

Many suppliers are inching towards this consolidated content management idea and are coming up with different application directions:

Looking across these six categories we can view them as converging into a single product/service category, which is a multi-type service including content capture, storage, hygienic cleansing, access, continuity, search, eDiscovery, retention and compliance - a super content management system that will gradually be developed over the next five to ten years.

Convergence

According to Mimecast's Murray, it is easier, (meaning more cost-effective,) and simpler to write the code for this from scratch and do it on a Google grid-like system infrastructure. It is better, he says, to do it this way and grow a core infrastructure that can scale, rather than buy in technology and cobble it together with existing products to provide the comprehensive functionality required. It is also better to do this on a massively scalable grid server base because that is far more cost-effective than using existing server-file storage array combinations.

Only by parallelising functions across such a grid can you get the responsiveness needed and on a cost-effective scale without the expensive over-provisioning you would be forced to use with conventional server-filer technology to meet the same goals.

Mimecast's grid infrastructure application has much Java code, a grid-based relational database service, and the open source Lucene full text indexing application with added grid awareness. The Linux x86 commodity servers running this code are linked across gigabit Ethernet with SATA drive arrays storing the data.

Murray is convinced that this Mimecast-developed, Google-like infrastructure is far more cost-effective than any cobbled-together set of applications running on conventional server-filer infrastructure. By providing its software-as-a-service, Mimecast can spread its costs across many customers. By writing its own software, profit margins are enhanced.

The company is making progress, gaining customers and even capturing customers moving away from MessageLabs and WebSense, according to Murray.

Mimecast offers a Time Machine for business capability, it being able to respond to requests to restore Joe.business's MyBusiness directory on date and time X. All the content is stored with data and time and more meta-data enabling this kind of restoration, and even reconstruction, to be carried out.

Filer supplier dilemma

Here's the filer supplier dilemma coming.

Murray says Mimecast provides three key email benefits: archiving, hygiene and continuity. These are being extended to cover general files, then SharePoint and, no doubt, even more content types in future.

As customers are handing their data over to Mimecast they are thinking about, and being asked about, backing up that content. Mimecast protects it for them, offering business continuity and disaster recovery services for the data, so customers don't need to do it themselves.

Mimecast extends its content type to general documents and more and more of this customer data gets handed over.

Murray says: "We could have end users mapping files to our drives [in the cloud]", to an M: drive, a logical drive built and presented from Mimecast's grid infrastructure. He goes on: "We round up the docs and then we're in a very strong position to say to business, stop backing up." Mimecast will already have the data and it doesn't need protecting further, customers will no longer need to back it up to virtual tape libraries or to real tape libraries.

In fact, with his view, customers no longer even have to store it themselves. They don't need to buy file storage arrays that store the data, snapshot it, replicate it to remote sites, and deduplicate it. A whole file-based system infrastructure becomes redundant as - if - a Nick Carr-type Big Switch happens and content storage and processing moves into the cloud.

What does this mean for existing suppliers?

Multi-play suppliers, such as backup vendors like Symantec and EMC, are moving to add cloud backup services. The road is open for EMC to combine its Mozy cloud backup and backup software with its Documentum ECM offerings into an aggregated cloud content management (capture, protection, access, search, etc.) service.

But Murray thinks that pure play vendors in this aggregating space will be in trouble as customers hand over the multifarious parts of their existing infrastructure to cloud content management suppliers - which is what Mimecast is becoming. A filer supplier currently offering storage arrays which snapshot, replicate, RAID protect, deduplicate and/or offer VTL functions will find its storage management function IP is no longer needed. The 10-year trend that Murray is seeing here leads inevitably, if he is right, to the destruction of the filer business.

Of course, this means that you have to take the view that private clouds won't happen and that a mass move of data into external clouds won't happen either. Today's cloud hype is just as fatuous as previous ASP (Application Service Provider) expectations.

Of course, again and in turn, this means you have to say that Salesforce.com, Google, Amazon and other cloud software service providers are not Big Switch harbingers at all, merely one-off wonders. Which way do you think things will go? If you are a product or service supplier to customers in this area, which way are you going to bet your company? ®

Download Mimecast whitepapers from The Register's research library