Sizing

##Sizing guideline for the Dynatrace AppMon & UEM SaaS Offering##

Choose deployment scenario

Two deployment types are supported for SaaS environments: 

Make sure to review both supported scenarios to understand the architecture, advantage & constraints linked with both options.

Fill Deployment Sizing Calculator

The Deployment Sizing Calculator is a spreadsheet you can use to calculate the recommended deployment size and categorize the result as a small, medium, large, or extra-large (XLarge) deployment, all based on sizing data you provide, such as the number of JVMs, web servers, and daily transactions. In addition, it calculates the recommended bandwidth, Session Storage, and Performance Warehouse sizing.  The Dynatrace AppMon & UEM SaaS managed server offering is available for Small, Medium and Large deployment. X-Large deployment is not part of the SaaS offering.
The SaaS Deployment Size matches the On-Premises Deployment Size (Small on SaaS = Small On-Prem, Medium on SaaS = Medium On-Prem & Large on SaaS = Large On-Prem). 

UEM + agents deployment

Deployment Size Session Storage Memory available
on EC2 host
Virtual Core Typical Network Bandwidth required to the Server in the cloud (see also below Network / Bandwidth below) Performance Warehouse
(initial; growth on-demand)
Analysis Server
Small
10 Agents | 300 Transactions/sec*
250 GB  7.5 GiB 4 0.75 Mbps
(2.5 Mbps uncompressed)
20 GB No
Medium
250 Agents | 1,500 Transactions/sec*
1024 GB 30 GiB 16 3.5 Mbps
(14 Mbps uncompressed)
40 GB On Demand / Depending on the server load
Large
750 Agents | 3,500 Transactions/sec*
2048 GB  60 GiB 32 8 Mbps
(32 Mbps uncompressed)
60 GB On Demand / Depending on the server load

* with average PurePath length of 100 nodes

UEM Only

Deployment Size Session Storage Memory available
on EC2 host
Virtual Core Performance Warehouse
(initial; growth on-demand)
Small
20 million visits/month
240 million visits
/year
250 GB  7.5 GiB 4 20 GB
Medium
40 million visits/month
480 million visits
/year
1024 GB 30 GiB 16 40 GB
Large
80 million visits/month
960 million visits
/year
2048 GB  60 GiB 32 60 GB

* with 15 user actions per visit and 10 nodes per user action PurePath

Review additional bottlenecks

Network/bandwidth

Experience shows that the usual bottleneck in high traffic environments is the bandwidth that’s available between the customer environment (Collectors installed on-premises) and the Dynatrace SaaS instance in the cloud. The Deployment Sizing Calculator provides an indication about the expected bandwidth requirement - however network bandwidth is difficult to estimate because it heavily depends on the type of instrumentation. For example, large attachments such as stack traces collected with exceptions or SQL statements cause much more traffic than simple PurePath nodes. In typical scenarios, every 500 TPS sent from a Collector to a Dynatrace Server will contribute 8 Mbps (1 MBps) of upstream bandwidth. The downstream is negligible in comparison. 

  • See Network Bandwidth Guidelines for more information on typical bandwidth requirements between Collectors and Server
  • The jPerf Server is installed per default on all the SaaS instances; contact the support team to have it enabled to test the bandwidth with jPerf as described here: Network Bandwidth Best Practices

Storage

  • Session Storage space is provisioned according to the deployment size (see above). 
  • Performance Warehouse (database) space starts with an initial storage size depending on the deployment size (see above). The database growth is monitored and the space is extended once 90% of the space available has been used.