How many people can use a single Hadoop cluster at one time?

I would like to know how many people can use a single Hadoop cluster at one time? I am asking because I need to figure out whether or not a single 5 or 10 node cluster would be sufficient to host a class of 12 to 24 students.

Also, I am wondering if anyone could recommend whether the specifications each for the nodes on a high-end educational-level cluster should be the same as for any other production-level cluster ( ie. 64G-128G RAM, 24TB hard drive space, 8 cores, etc.). I believe that the dataset sizes that students would be using will range between 20MB minimum to about 0.500TB maximum, I imagine that we will ultimately be working on real problems and datasets, even if they are not exactly considered to be big data.

Topic data apache-hadoop education

Category Data Science

1
answered at

The bottleneck depends on the use pattern rather than the direct number of users. If people are doing high I/O workloads, then you wont get many people on at all. Whereas if you are doing small processes or just using it as a simple data lake, you could host the sort of people you're talking about there.

Have you considered using AWS?

1
answered at

For initial learning you could very easily do proof of concept work against individual VMs (a 4GB VM with a pseudo cluster is enough to do basic mapreduce examples in). If you're going to use Spark I would lean towards higher memory counts if they're within budget. I'd also keep an eye towards more, lower cost nodes. A stack of Intel i7 NUCs (or similar) with 2TB consumer SSDs and 32GB of RAM will cost less than $1k per node and 10-15 of them could easily handle a class of 24 students.

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.