Internet Scale Databases

Rob and I have just published a presentation on slideshare titled Mumps the Internet Scale Database.

It looks at the fit between the plain old Mumps database and the needs of Internet scale applications (think Google search, Twitter, Facebook, eBay, etc).

Many of the largest scale Internet plays have found conventional relational databases to be inadequate for their needs. In most cases they've had to innovate and cobble together something useable from scratch. The traditional SQL solutions were neither scaleable enough or affordable. Some of their solutions, such as Amazon's SimpleDB and Google's BigTable, are now starting to appear as commercial products that others with Internet-scale needs can also use.

I've written previously about Drizzle, a fork of MySQL, which tries to be an Internet-scale solution, by throwing out a lot of SQL clutter. In most cases, when trying to create a database that will scale sufficiently, the first casualty is SQL. This is certainly the case for BigTable and SimpleDB. Neither have full implementations of SQL and we get back to schemaless, hierarchical databases. Why? Because they are fast and flexible. The overheads of SQL add nothing in the context of large scale Internet databases and simply slow down the speed of development and, more importantly, the performance of the application.

Our presentation shows how plain old Mumps is a well proven database solution with a track record of success that meets all of the needs of Internet scale applications. It explains what the issues are and how the new kids, Amazon and Google, have had to innovate to scale. It goes on to show how Mumps has already solved these problems and is a freely available alternative to Amazon's and Google's pay-per-use propositions.



Part of the point of my recent presentation about BigTable was that conventional databases just don't cut it for Internet Scale applications.

Drizzle is newly announced fork of MySQL that cuts out a lot of features (triggers, stored procedures, etc) with the goal of producing a highly scalable database that is suitable for Internet Scale problems.



The Quest for an ObjectScript Iterator

I've been following an interesting blog by someone who is finding their way around Cache for the first time: http://cachetastic.blogspot.com/

I've added it to my feed reader.

Old Relics

Colossus, arguably the first digital computer, was created at Bletchley Park during WWII. In keeping with the spirit of this, we are inviting everyone to bring along items of old computing capability. If you have any old bits of computer equipment (early PCs, adding machines) or related relics (log tables, sliderules, magnetic tape, punched cards, etc) hidden away in your attic or garage, please bring them along. There will be a small display area where everyone will be able to view and discuss these items during the day.

Syndicate content