4e Distributed Databases +/-

4e Distributed Databases +/-

1. Storage

   • Partitioned
     • O-Less storage required at central location
     • X- local server need their own storage
     • X- Increases over-all storage needed
   • Replicated
     • X- lots more storage space needed
     • X- Each location needs large expensive hard disks and complex set up
   • Centralised
     • O- Storage only needed at central location
     • X- Can be expensive as storage need to be high quality to allows multi-user access and powerful processors will be needed at the central location.
     • O- local server only need to store small indexes

2. Database Design

   • Partitioned
     • X- Very complex. Requires trained specialists to create and maitain
   • Replicated
   • Centralised
     • O- design is straight forward, following that of a relational database

3. Local Control

   • Partitioned
     • O- Each site in control of its' data and can customise its database to suit them.
     • X- If customisation is not controlled, the data may become unavailable or in wrong format for other sites
   • Replicated
     • O- Replica can be customised o be more usefue.g. adding extra fields
   • Centralised
     • X- Local sites have no control over the database so cannot customise it to their needs
     • O- Only one configuration is maintained, which maintains data integrity

4. Queries

   • Partitioned
     • O- local queries are faster as there is less data to search through
     • X- Global queries are much slower as data will need to be searched and returned from many different sites
   • Replicated
     • O- Queries are quicker as no requests needs to be sent to other sites
     • O- No difference between global and local queries
   • Centralised
     • X- Searching is as fast as external communication link allows between local and central site
     • X- Without an index processing will have to be done at the central server which may be busy with other tasks
     • O- With an index processing is quickly done by local server and request for the correct records is sent.

5. Data Availability

   • Partitioned
     • O- If one data link breaks, other sites will still be able to get data from each other. The affected site will still be able to process local data
   • Replicated
     • O- If link is broken each site can continue working with its own replica
     • X- Updates cannot be done with a broken data link
     • X- Updates can only be done so often to reduce network traffic
       • Not suitable for real time data
   • Centralised
     • X- If data link to central location is broken, the site can't have any access to the database
     • X- If the server at the central location breaks down, no-one can access the database anywhere
     • O- Data is always up-to-date. No conflicts that can compromise data integrity.
       • Good for real time data

6. Back Ups

   • Partitioned
     • X- Back-ups can't be taken centrally as data is fragmented
     • X- Each site needs it's own back-up routine
   • Replicated
     • O- Back ups only needed if whole system becomes corrupted
     • O- Each replica acts as a back-up
   • Centralised
     • X- A fail safe back-up system is needed
       • Needs to included changing tapes daily and tape rotations
         • Back-ups must be tested that they can be restored and stored in an off-site location.