I have done an Enterprise Synthetic deployment at a customer location, the customer is looking for a sizing calculation some thing close to what we have for Dynatrace App Mon, they would like to know how many transactions can be done on a single agent, how many agent connection can be done to the reporting server (reporting server contains CSS, CAS, Agent Manager, RUM Console and Enterprise Console) and what is the impact on the resources with respect to CPU, Memory Network etc.. on the reporting server and Agent
i did go through the documentation and just found some thing for database
"Database sizing is highly dependent on your environment's configuration and monitoring requirements. As a general guideline for estimating appropriate database size, consider that for every 2,000 transactions, the Enterprise Synthetic database grows on average approximately 1 MB.
For example, 20 Agents running 1 transaction each every 12 minutes generates 100 transactions per hour (2,400 transaction per day) causing the database to grow by 1.2 MB per day."
I will need some thing close to the Database Sizing example above for Reporting server components and Agent Components
Any help in this regard will be great
Answer by Frans S. ·
Following your question with great interest. Agent Capacity management is hot topic here.
I think your question falls out into two subjects: Server side, and Agent side capacity.
Regarding the Agent:
Actually the Agent Transaction (load) capacity depends on various factors, besides the hardware capacity itself. Such as what type of transaction, autocheck, automated (recorder) or third party.
Most importantly: The duration (in time) of a task (script) (the time it takes, from start to finish). (The more powerful the agent the shorter the duration). A Recorder script (Task) hold one or more transactions.
Regarding automated (recorder) transactions, where only one can be active at the time, the sum of the duration of all tasks (assigned transactions * interval) determines if an agent is able to process all these tasks, and thus transactions.
A synthetic agent can cary out just so many tasks in one hour. So the capacity is 60 minutes... So you actually have to, up for hand, baseline each task (Recorder script), to the expected amount of time. Then determine the desired Interval (frequency). (e.g. A Task that runs 1 minute every 10 minutes, consumes 10 minutes of the hour. 60-10). And with that calculate the resource impact.
You also have to specify a capacity margin. Due to issues/timeouts a transaction can take longer or time out. So you can not fully utilize the 60 minutes of the hour. We use 30 minutes for actual task execution per robot as capacity margin.
Then comes the actual capacity management afterwards. Currently we think the product lacks support for this. Therefor with a combination of a database query, and an Excel sheet, we do capacity planning on customers agents.
For autochecks, that run in background, I think applies another rule of thumb.
Order of Script Execution on same Robot? 4 Answers