Equivalent droplet sizes to EC2 instances

May 14, 2013 6k views
Hello, I've been trying to put out a table that compare some EC2 instance sizes we use for one of our applications and the equivalent in DO droplet sizes. For example, we use the following sizes: High CPU Medium (32bits) - c1.medium 5 ECU, 1.7GB RAM High CPU Extra Large (64bits) - c1.xlarge 20 ECU, 7GB RAM Large (64bits) - m1.large 4 ECU, 7GB RAM For example, 20 ECU High CPU are used for DB purposes while 5 ECU are for smaller web FE (and sometimes we use High CPU Extra Large too) What you think will be the equivalent of those for instances? Thank you
1 Answer
That actually depends on how much storage offered by Amazon you are using, because you can match both the CPU and RAM sizes of AWS and DO (DO comes oout a bit better), but the storage is limited on DO, while it's virtually unlimited on AWS plus you have the ephemeral storage attached to the instance there as well. Plus there is the networking (see below) For example:

c1.medium can be replaced by the $20/month DO droplet, you have the same amount of 2 cores with 2GB RAM. CPU speed would rougly be the same (run some tests or compare on serverbear). What you lose is the HDD. You only get 40GB with the DO droplet, whereas c1.medium will give you 350GB ephemeral plus whatever you allocate as EBS storage (for additional costs of course). Then you have pricing, where DO only costs you $20/month, while c1.medium will cost you as on-demand instance in the US East 1 region cca. $105/month plus EBS storage costs plus traffic costs.

Same can be done for the others:

c1.xlarge being the DO 8 core 16GB RAM droplet with 160GB storage. Costs are $423 for c1.xlarge versus $160 for DO. This is to match the core count, you get double the RAM. Or you could go with the 4 core 8GB droplet for $80 per month.

m1.large being the 4 core 8GB RAM droplet with 80GB storage. Costs would be $175 versus $80 for DO.

What's important here as well is that with you don't have all the functions AWS offers, so for example networking will be an issue because DO does not seem to have an internal network yet for communicating between the droplets. This may be a total deal breaker for you even though the instance/droplet pricing itself is heavily in favor of DO if you fit into the offered storage.

Have another answer? Share your knowledge.