Elasticsearch 1.6.x to 2.3.x migration

Elastic Stack Elasticsearch
1

Hi,

We have been using elasticsearch as a data-store for our application.

In the application, we were storing various devices and their outputs values.
For storing all the device's details like name or other parameters, we have a "type" called master.
And all the device output values are stored in "type" which is same as the device-name. This was done for faster searching.

When we used to delete a device, we used the delete the device from the master type and delete the "type" with the same device name.

But with the new version of elasticseach, we are not able to delete the type anymore. We can delete all the documents in the type.

So, will many unused type create a problem with huge data-set (700GB+)? Will it affect the performance?

Is there any work-around for this??

Thanks in advance

2

The best option is to split things into their own indices.

You may want to reindex to 2.X anyway, to take advantage of doc values.

3

In the application, a user is an index and all its devices types. some users have more than 10000 devices. Having so many indices won't effect the performance???

4

Ah ok.

Well you have a choice to make, use the delete by query plugin, use multiple indices, or just leave the data and then delete it using your higher level retention policy.