This page describes hardware and software requirements for the Controller hosted on private or public cloud to help you prepare for your AppDynamics deployment.
About Controller Sizing Information
Every deployment is unique. Factors such as the nature of the application, workload, and the AppDynamics configuration can all affect the resources required for your specific scenario. Be sure to test the performance of your system in a staging environment, so that you can fully understand your requirements before deploying AppDynamics to its live operating environment.
Before installation, it's usually easiest to estimate your deployment size based on the number of nodes. For Java, for example, a node corresponds to a JVM. However, the best indicator of the actual workload on your Controller is provided by the metric ingestion rate.
After initial installation, you should verify your Controller sizing using the metric upload rate. You then need to continue to monitor the Controller for changing workload brought about by changes in the monitored application, its usage patterns, or in the AppDynamics configuration.
General Hardware Requirements
The following general requirements that apply to the machine on which you install the Controller:
- The Controller should run on a dedicated machine. A production Controller must run on a dedicated machine. The requirements here assume that no other major processes are running on the machine where the Controller is installed, including no other Controllers.
- The Controller is not supported on machines that use Power Architecture processors, including PowerPC processors. The Controller is supported on amd64 / x86-64 architectures.
- Ensure that the Controller host has approximately 200 MB of free space available in the system temporary directory.
- Disk I/O is a key element to Controller performance, particularly low latency. See Disk I/O Requirements for more information.
The following table shows Controller installation profiles by metric ingestion rate and node count. As previously noted, the actual metrics generated by a node can vary greatly depending on the nature of the application on the node and the AppDynamics configuration. Be sure to validate your sizing against the metric ingestion rate before deploying to production.
General Sizing for On-Premises
|Profile||Max Metrics/Minute||Max Agents (Approx)||OS||Compute||Storage|
|Demo||10,000||5||Linux or Windows||2 Cores, 8 GB RAM||50 GB|
Note: This profile is not supported when installing with Aurora DB.
|Small||50,000||50||Linux or Windows||4 Cores, 16 GB RAM||400 GB|
Note: This profile is not supported when installing with Aurora DB.
|Medium||1,000,000||1,500||Linux or Windows||Bare-metal: 8 Cores, 128GB RAM||5 TB SAS SSDs. Hardware-based RAID 5 configuration|
|Linux or Windows||VM: 16 vCPUs, 128GB RAM||5 TB SSD-based SAN with 10 Gb/s FCoE|
|Large||5,000,000||10,000||Linux||Bare-metal: 28 cores, 512 GB RAM|
|Extra Large||For deployments that exceed the Large profile configuration defined here, contact AppDynamics Professional Services for a thorough viability evaluation of your Controller.|
Amazon Web Services (AWS) Sizing for On-Premises
|AWS Profile with Aurora||Max Metrics/Minute||Max Agents (Approx)||OS||Compute||Instance Size for AWS Aurora Storage||Block Storage (for Controller application files only)*|
|Medium||1,000,000||1500||Linux||EC2: r4.2xlarge||db.r4.4xlarge||10 GB GP2 EBS Volume. We recommend using a different volume than the instance's root volume.|
|Large||5,000,000||10000||Linux||EC2: r4.8xlarge||db.r4.16xlarge||10 GB GP2 EBS Volume. We recommend using a different volume than the instance's root volume.|
* The specified disk space must be available for use by the Controller. Specifications do not include overhead from the operating system, file system, and so on.
Elastic Network Interface (ENI)
The ENI numbers were last updated on Feb 28, 2018.
For AWS, provision an ENI for each Controller host and link the license to the MAC address of the ENI. For more information about ENI, see the AWS documentation at the following link:
Additional Sizing Considerations
Note the following additional requirements:
- Large installations are not supported on virtual machines or systems that use network-attached storage.
- The RAM recommendations leave room for operating system processes. However, the recommendations assume that no other memory intensive applications are running on the same machine. While the Enterprise Console can run on the same host as the Controller in small or demo profile Controllers, it is not recommended for medium and larger profiles or for high availability deployments. Refer to the Enterprise Console Requirements if the Enterprise Console is on the same host as the Controller.
- Disk sizing shown in the sizing table represents the approximate space consumption for metrics, about 7 MB for each metric per minute.
- The motherboard should not have more than 2 sockets.
- See Calculating Node Count in .NET Environments for information related to sizing a .NET environment.
The agent counts do not reflect additional requirements for EUM or Database Visibility. See the following sections for more information.
Disk I/O Requirements
A critical factor in a machine's ability to support the performance requirements of a Controller in a production environment is the machine's disk I/O performance.
There are two requirements related to I/O latency:
- This disk I/O must perform such that the maximum write latency for the Controller’s primary storage must not exceed 3 milliseconds while the Controller is under sustained load. AppDynamics cannot provide support for Controller problems resulting from excessive disk latency.
- Self-monitoring must be set up for the Controller. Self-monitoring consists of a SIM agent that measures the latency of data partitions on the Controller host, and the configuration needs to include dashboard and health rule alerts that trigger when the maximum latency exceeds 3 ms. For details on Controller self-monitoring, contact your AppDynamics account representative.
Disk I/O Operations
The AppDynamics Controller performs two types of I/O operations important to Controller performance:
- The MySQL intent log is very sensitive to latency, and MySQL performs writes using varying block sizes.
- MySQL’s InnoDB storage engine uses random, asynchronous, 16kB reads and writes to move database pages between storage and cache. In a properly sized Controller, most reads are satisfied from one of the software caches.
It’s important for best performance that the stripe size of the RAID configuration matches the write size. The two write sizes are 16Kb (for the database) and 128Kb (for the logs). You should use the smallest stripe size supported, but no smaller than 16Kb. If using a hardware-based RAID controller, be sure that it supports these stripe sizes. The stripe size can be determined by the number of data disks multiplied by the strip/segment/chunk (the portion of data stored on a single disk).
SAN-based Storage Limitations
While onboard disks typically satisfy I/O requirements, SAN-based storage could be hampered by poor I/O latency performance. In addition, AppDynamics discourages the use of an NFS-mounted filesystem. NFS adds latency and throughput constraints that can negatively affect Controller performance and even lead to data corruption. Similarly, you should avoid iSCSI or other SAN technologies that are subject to quality of service issues from the underlying network.
If you choose to deploy one of these latency-challenged storage technologies on a system that is expected to process 1M metrics/min or greater, a mirrored NVMe configured as a write-back cache for all storage accesses is recommended. Configuring such a device will hide some of the longer latencies that have been seen in these environments.
In all cases, be sure to thoroughly test the deployment with real-world traffic load before putting an AppDynamics Controller into a live environment.
End User Monitoring (EUM) Considerations
End User Monitoring (EUM) typically increases the number of metrics collected. Accordingly, the Small Controller profile is not supported for installations that use EUM. A Medium profile running 20+ high-traffic BRUM/MRUM agents should be sized at a specification closer to a Large profile for EUM.
Specifically, EUM impact metrics as follows:
- Web RUM can increase the number of individual metric data points per minute by up to 22000
Mobile RUM can increase the number of individual metric data points per minute by as much as 15 to 25K per instrumented application if your applications are heavily accessed. The actual number depends on how many network requests your applications receive.
- Monitoring EUM is memory intensive and may require more space allocated to the metrics cache.
The number of separate EUM metric names saved in the Controller database can be larger than the kinds of individual data points saved. For example, a metric name for a metric for iOS 5 might still be in the database even if all your users have migrated away from iOS 5. So the metric name would no longer have an impact on resource utilization, but it would count against the default limit in the Controller for metric names per application. The default limit for names is 200,000 for Browser RUM and 100,000 for Mobile RUM.
Database Monitoring Considerations
The following guidelines can help you determine additional disk and RAM required for the machine hosting the Controller that is monitoring the Database Agent. For very large installations, you should work with your AppDynamics representative for additional guidelines.
For on-premises installations, the machine running the Controller and Event Service will require the following additional considerations, for a data retention period of 10 days:
- 1 – 10 collectors: 2 GB RAM, Single CPU
- 10 – 20 collectors: 4 GB RAM, 2 CPUs
- More than 20 collectors: 8 GB RAM, 4 CPUs
Sizing the Controller for the Events Service
The Events Service is a file-based storage facility used by EUM, Database Monitoring, and Analytics. Database Monitoring uses the Events Service instance embedded in the Controller by default. The disk space required will vary depending upon how active the databases are and how many are being monitored.
For redundancy and optimum performance, the Events Service should run on a separate machine. For details on sizing considerations, see Events Service Requirements.
Calculating Node Count in .NET Environments
The .NET Agent dynamically creates nodes depending on the monitored application's configuration in the IIS server. An IIS server can create multiple instances of each monitored IIS application. For every instance, the .NET Agent creates a node. For example, if an IIS application has five instances, the .NET Agent will create five nodes, one for each instance.
The maximum number of instances of a particular IIS application is determined by the number of worker processes configured for its application pool, as illustrated in the following diagram:
The diagram shows three application pools — AppPool-1, AppPool-2, and AppPool-3 — with the following characteristics:
- AppPool-1 and AppPool-3 can have a maximum of two worker processes (known as a web garden), containing two applications (AppA, AppB) and one application (AppF), respectively.
- AppPool-2 can have one worker process. It has three applications.
To determine the number of nodes, for each AppPool, multiply the number of applications by the maximum number of worker processes. Add those together, as well as a node for the Windows service or standalone application processes.
The example would result in nine AppPool nodes. Adding one for a Windows service would result in a total of ten nodes, calculated as follows:
To find the number of CLRs that will be launched for a particular .NET Application/App Pool:
- Open the IIS manager and see the number of applications assigned to that AppPool.
- Check if any AppPools are configured to run as a Web Garden. This would be a multiplier for the number of .NET nodes coming from this AppPool as described above.
Asynchronous Call Monitoring Considerations
The Small profile is not supported for installations with extensive async monitoring. A Medium profile running 40+ agents may need to upgrade to a configuration closer to a Large profile if extensive async monitoring is added.
Specifically, monitoring asynchronous calls increases the number of metrics per minute to a maximum number of 23000 per minute.