In February 2012 I wrote about “Environmental Spatial Data Infrastructure” on this blog. Later that year the case complex matured somewhat and in August I wrote the posting “Clearinghouse for the environment – the scaffolding (I)“.
Since then I have together with my colleagues had the opportunity to test systems in full scale by contributing to the implementation of clearinghouses in partner countries.
Last time I showed how a stack consisting a hardware layer with vmWare as one of the basic modules could form the basis of an environmental spatial data infrastructure. In some ways this was a rather optimistic setup.
Some of our main challenges with the above set up was maintenance of physical equipment. So we removed that layer. Maintaining a complex setup with a virtual machine environment, or getting access to local environments proved in general to be difficult. So we ditched it.
To get the systems running we needed:
- Shared access to the systems for administrative purposes
- A flexible backup-system
- An option to duplicate successful setups
- High availability
- Flexible security system
Other things we considered important
- The system should not tie our partners up in future licensing costs
- Compliance to central standards
- An option for partners to move the systems to physical infrastructure if necessary
- Option to keep traffick outside our own company networks – since they are de-facto external systems paid for by external partners
As you all can see in all a lot of considerations which we had to relate to.
Since internet access across borders in any case would be relevant for retrieving external map layers we started looking at how we could use Amazon services. I already had experience in running virtual macines using Amazon EC2. Amazon helped us out with many of the issues mentioned above. So in short we moved the whole setup to Amazon. The following figure illustrates the setup.
In addition to the components relying on EC2 we have also found that using Amazon Simple Storage (S3) for storing survey data of some size could be a good ide. S3 allows the user to distribute files using “secure” links and even using the bit-torrent protocol for files up to 5 Gb.
We now have one such system built up and under testing. It looks good but as always the technology is but a small part of the equation. Establishing information flows, using standards etc represents the major parts of a national envoronmental spatial data infrastructure.
Should the need arise to develop custom made solutions it should be possible to add more virtual machines in the setup.
Given that our partners find this setup trustworthy we will probably suggest this as an entry level spatial data infrastructure for environmental data.
WordPress directly, and through countless plugins, supports many standards for embedding information. How information should flow between the different systems in this setup has been given some thought. I will try to elaborate on this in a later posting – hopefully in less than two years time.