Cloud Server Recommendations

Dear all,

We are working on a large-scale survey targeting 1,000 firms. The data collection process will begin in September and we want to use Survey Solutions to design the survey and collect the data. As we do not have access to an existing server to store the data, we have been looking into cloud service providers such as AWS for storage.

We would like some guidance/recommendations on which type of server and which instance would be most suitable for our type of project. For example, we have been looking into Amazon EC2 as an option, but we do not know which instance would be best. We would like the server to be accessible both by Mac and Windows laptops. We will not collect any multimedia data, we only plan to collect quantitative data and the GPS coordinates of the respondents.

For AWS: Would the Storage Optimized Instances be suitable for the project? If so, which features should we be looking at when comparing the different instance types (Im4gn, Is4gen, I5i, I3, I3en, D2, D3, D3en, H1). If we are planning to collect data on 1,000 firms which vCPU and Memory (GIB) would we need?

We are also open to suggestions on other cloud server providers as well. Thank you so much for your help!

Best,
Elena

Dear Elena,

Form my experience the process that demands the most resources from the server is the sync process with the tablets. If your plan is to use tablets to collect the interview. You have to pay attention to the number of interviewers synchronizing to the server and the number of interviews collected per day.

We just finished a project where we collected 13,000 interviews with a team of 200 interviewers. For this projects we decided to host the server in AMAZON EC2 , we us a t3.Xlarge instance based on windows . The platform cost an average of $190 per month.

How many interviews you plan to collect ? How many interviewers will work on the field ?

Best

Vladimir

Dear Vladimir,

Thanks for your reply and for sharing your experience! We plan to collect 1,000 completed interviews in around 2-3 months. We will have around 25-30 enumerators on the field synchronizing the interviews with tablets. We expect (at most) 1 interview per enumerator per day.

I see that the t3.Xlarge instance is a General Purpose instance. Would you recommend choosing one of the General Purpose instances rather than a Storage Optimized instance? Perhaps a t3.medium would be enough for us (2 vCPU, 36 CPU Credits/hours, 4 Mem (GiB))?

Thank you so much for your help!

Best,
Elena

Hi Elena,

For our project we decided to follow the recommendation of this article in SuSo KB (Server Requirements) 16GB ram and 4 CPU.

I agree with you, the t3.medium (4Gb ram and 2 CPU) instance would be enough because the workforce for you project is small (25-30 interviewers).

Regards

Vladimir