Hardware for Large Aternity Deployment (up to 110,000 devices)

The large Aternity deployment supports up to 110,000 monitored devices or virtual desktops, where up to 100,000 of them are physical devices and 10,000 are virtual application sessions. It assumes 10 Aternity logged in users updating dashboards and performing in total up to 10 clicks per minute.

Large Aternity deployment

Aternity requires one Aggregation Server per 40,000 monitored devices and one Vertica Database Server node per 25,000 monitored devices. So, calculate the number of servers you need following the amount of monitored devices in your organization.

Data Warehouse Server should be deployed on a dedicated machine separately from other Aternity servers in deployments with more than 40,000 monitored devices.

The disk space is for each separate Vertica Database Server node in a cluster, before any high availability (k-safety = 0). For minimum high availability, we recommend k-safety = 1 (learn more).

Hardware Specifications for Large Deployments

The large deployment requires the following hardware specifications:

Component Minimum Memory (RAM) Minimum CPU Minimum free disk space Additional configuration

3 x Aternity Aggregation Server

12GB

Xeon E3 family, 2 cores at 2.4 GHz

30GB

HEAP_SIZE= (learn more)

-Xmx8192m

-Xms8192m

-XX:NewSize=4096m

-XX:MaxNewSize=4096m

Aternity Management Server

At least 22GB

Xeon E5 family

4 cores

2.4 GHz

80GB

HEAP_SIZE= (learn more)

-Xmx14336m

-Xms14336m

-XX:NewSize=8192m

-XX:MaxNewSize=8192m

ActiveMQ JVM Heap Size: 2G

Aternity Dashboard Server

64GB (you can ignore the RAM size warning during setup)

Add 16GB RAM for each additional 10 logged users.

Xeon E5 family

8 Cores

2.4-3.6 GHz

15MB CPU cache

Add 8 cores for each additional 10 logged users.

80GB (SSD) on a dedicated drive

Or I/O per sec (IOPS): 900

For less standard hardware configurations, see the setup instructions and hardware best practices from Tableau. To deploy on a virtual machine, you must guarantee the CPU and memory requirements.

Aternity Data Warehouse Server

At least 22GB

Xeon E5 family

4 cores

2.4 GHz

50GB on a dedicated drive

HEAP_SIZE= (learn more)

-Xmx16384m

-Xms16384m

-XX:NewSize=8192m

-XX:MaxNewSize=8192m

Aternity Oracle Database Server

32GB

Xeon E5 Family

8 Cores

2.4 GHz

600GB

Or I/O per sec (IOPS): 1200

Oracle's System Global Area (SGA): 20GB

Program Global Area (PGA): 8GB

5 x Aternity Vertica Database Server

64GB

Xeon E5 family

8 Cores

2.4-3.6 GHz

15MB CPU cache

560GB (SSD) for each node

  • The disk space is for each separate Vertica Database Server node in a cluster, before any high availability (k-safety = 0). For minimum high availability, we recommend k-safety = 1 (learn more)
  • If you deploy on virtual machines, you MUST define all resources as dedicated or reserved (Resource Allocation > Reservation). In addition, you can use a Logical Volume Manager (LVM) in Aternity 12 on any hard drive with no restrictions. See Vertica limitations for LVM.

  • Verify that the disk partition dedicated to the Vertica data folder (Target directory) uses the EXT4 file system. Enter df -Th to verify.

  • It is recommended to utilize less than 60% of this partition, allowing high availability operations to run properly. If utilization rises above 60%, increase volume size to compensate

  • Network – requires at least 1Gb/s interface, ( recommended 10Gb/s).

  • Disk IO – 300 IOPS per 100GB storage (for 780GB > 2400 IOPS) and 80MB/s read/write throughput per core (for 8 cores > 640 MB/s read/write throughput).

Aternity Docker Components Server

32GB

Xeon E5 family

16 Cores

2.4-3.6 GHz

620GB

20GB for partition with Docker engine’s local storage, by default located at /var/lib/docker/.

Divide free disk space into several partitions where each partition is dedicated to the data directory of a different component: Messaging Broker(Kafka), and Raw Data Component (Cassandra). Allocate disk space to each partition according to hardware requirements of the relevant sizing model. The rest free disk space is dedicated to log files and REST APIs.

<cassandra>: at least 300GB

<kafka>: at least 150GB

If there is no need for optional components, you can reduce 6GB of memory and 1.5 CPU cores.

If you do not deploy Aternity Data Source for Portal, you can reduce the amount of memory by 2GB and CPU cores by 0.3.

If you do not deploy Aternity REST API Server, you can reduce the amount of memory by 2GB and CPU cores by 0.3.

You must also install each server with the required operating system and other software. For details, see the system and software requirements for each server.