Cluster Design question

I am setting up new cluster.

Will have few new systems. and have some question.

  • four high end servers with faster disk, 192gig ram, faster network connection with multiple cpu. (master/data)
  • one server will be little old on low end (master only)
  • one server will be low on storage space but 192gig ram, high network and faster cpu with memory (logstash)

  1. putting about 2Million record daily.,
  2. about 10 index will be busy all time constantly adding records.
  3. about 10 index will get some kind of data few hours or once a day not that much.
  4. Trying to keep each shard max 20gig for high end index with one replica
  5. all other lower size index will have one shard and one replica

Should I do this?
4x data/master - about 8TB total space for hot index, 6 tb for cold index
1x master only
1x logstash only

On logstash point all output to 4xdata/master ?
or should I spin up three container on logstash server that way I will have three logstash system?

for kibana
should I do same, get some old hardware and spin up some instance?

Any input?

am I on right track?