cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Dynatrace Managed storage Behaviour

ashutosh_singh1
Organizer

Hi All,

I would like to know about one agent data storage behavior, as appmon is storing statistical data in PWH and pure path related in session storage. Likewise one agent is storage is only cassandra or other storage also available.

if Yes, So how it bifurcate data to store in different types of databases.

Regards,
Ashutosh


3 REPLIES 3

Radu
Dynatrace Pro
Dynatrace Pro

Hi Ashutosh,

Firstly, in AppMon, the Performance Warehouse (separate db) is used to store time series and incident data while the purepaths were stored on disk as either a live session (expanding with as many purepaths as the quota allowed) or as exported sessions (and these were either purepaths, thread dumps, or memory dumps).

Now, Dynatrace
has three main types of storage, all located on the nodes of the cluster:

  1. Transaction storage (purepaths). This is the raw transaction data that we base
    our analysis on.
  2. ElasticSearch storage. This contains problem data and user sessions.
  3. Cassandra storage. This contains our time series data.

With one node, all the data is stored in one place. But when clusters have 3 or more nodes, ElasticSearch and Cassandra use replication so that each record of the data would be stored on 3 of the available nodes. Transaction storage, as far as I know, isn't replicated and so the transactions will be stored on the node which will process them (decided by the internal load balancing algorithm).

I hope this helps.

Best regards,

Radu


ashutosh_singh1
Organizer

Hi Radu,

Thanks for detail description...!

But i have few concern points(if we have cluster environment):
1. If one cluster server got down, so how will access transaction level detail of that server during that time period.
2. How we have to access detail information about all cluster transaction data, through one URL(which were shown after installation) or need to use different URL for different cluster servers.

Regards,
Ashutosh Kumar Singh


You have a Cluster Management Console (https://<managedURL>/cmc), an interface for managing your Dynatrace Cluster and the nodes, from there you can see in detail how much storage is available for each of these categories, and how much is used on each of the nodes.

If a node cannot be accessed, that node's data won't be available in the UI until it's up and running again. However, if the cluster has enough nodes, data collection and tool usage still continues despite one node being down.