Bump resources on metricsdb-01

We are getting a little wall with performances on metricsdb-01 and I was wondering if we could get more memory on it and maybe even cpu cycles to see if our queries run a little faster. Maybe we should also start separating metrics services from that machine. What do you think?