In our project, we have two specific classes of data that we want to store inside the redships cluster - one of the data There is a lot of space required for the class (terabytes), but we only run ad-related questions against it, so performance is not a big issue and another square of data is very small (ten gigabytes), but we have a big Number query Yes, you need to run the contrary, the performance is an important factor. Given that distinction - the first class of data is most suitable for HDD-supported nodes, while second-class SSD-powered is better on more performance oriented nodes. We can potentially separate this data into two separate groups, but I wonder if it is possible to mix different types of nodes in the same cluster and in some way control what nodes are on .
No, now you can not mix two types of nodes together. You have such a scenario as you have, so we have two clusters, which is a SSD with a low space and asks more questions, but the fastest and second is the non-SSD that is your bytes data used for historical reports. Is done where the query is not the execution issue.
No comments:
Post a Comment