Distributed Hash Map (DHT) research - setting evaluation criteria.
In this post I want to set up formal criteria to compare different DHT systems:
1) Queries support. It is very important topic in my opinion, since the complex queries is a main
advantage of the RDBMS solutions against DHT. The main things to check here I see
- lookup by the key, lookup by the attributes, lookup by the logical conditions over attribute. The way to emulate join will be a separate topic of research.
2) Transactions support. It is not exactly the black and white, since there are different parts of the ACID properties implemented with some limitations. Each property of ACID will be evaluated separately.
3) Performance. There is no industry standard benchmark for the DHT performance. So I will measure simple response
time for the available types of queries. It will not be regular benchmarking because I can not run it on the same hardware.
But if difference in orders of magnitude will be discovered it can give some hints what solution is faster.
If you see other important criteria to be evaluated – I will add them to the evaluation process.
No comments:
Post a Comment