Supported instances and Usecases

Menu

In order to cater to the increasingly broad range of applications, cloud providers offer a wide range of instance types. It is important to evaluate performance metrics by profiling, load testing and benchmarking. It takes months of effort to find the best suitable instance type in terms of performance and cost effectiveness for an application.

At Instaclustr, our engineers have studied and experimented with various cloud instance types to run Cassandra and Kafka clusters. We have undertaken several tests and benchmarking for different application workloads and have come up with a short list of instance types for Cassandra and Kafka to suit the application. Considering requirements for throughput, latency, data size and cost, we have put together the following table capturing the use cases where each of our instance types are best suited.

Apache Cassandra

AWS

Instance NameAWS Instance TypeMemoryCPU CoresStorage Capacity
EBS: tinym5.large8000 MB2250 GB (EBS SSD)
Use-Cases: Smallest available production node. Use this when getting started with minimal production traffic. When you need more capacity than a 3 node cluster, we recommend scaling up to xl nodes rather than scaling out with more m5.large instances.
 
 
EBS: small
m5.xlarge
16000 MB
4
400 GB (EBS SSD)
EBS: balanced800 GB (EBS SSD)
EBS: bulk storage1600 GB (EBS SSD)
High Memory: balanced
r5.xlarge
32000 MB
4
1200 GB (EBS SSD)
High Memory: bulk2000 GB (EBS SSD)
Use-Cases: This family provides a range of different ratio of disk to processing capacity to allow you to select an instance that best matches the profile of your workload. The additional memory of r5 instances can provide improved price/performance for many use cases and is generally our recommended starting point for a production deployment.
 
 
High Throughputc5d.2xlarge16000 MB8186 GB (Local SSD)
Maximum Densityi3en.xlarge32000 MB42328 GB (Local SSD)
Extra Largei3.2xlarge61000 MB81769 GB (Local SSD)
Use-Cases: Local SSD based instances can offer the best price-performance when your use case fits well into the available disk/CPU ratios. Where the application workload requires very low-latency I/O in addition to high throughput, these instances are more suitable than the r5 range as these come with local SSDs. The i3en.xl model can also offer the lowest cost storage option for some scenarios.
– c5d.2xlarge: high throughput, low data storage requirement
– i3en.xlarge: medium throughput, high data storage
– i3.2xlarge: balanced throughput/storage
 
 
Resizable Large
r4.large15250 MB23200 GB (EBS SSD)
r4.xlarge30500 MB43200 GB (EBS SSD)
r4.2xlarge61000 MB83200 GB (EBS SSD)
r4.4xlarge122000 MB163200 GB (EBS SSD)
r5.large1600 MB23200 GB (EBS SSD
r5.xlarge32000 MB43200 GB (EBS SSD)
r5.2xlarge64000 MB83200 GB (EBS SSD)
r5.4xlarge128000 MB163200 GB (EBS SSD)
Resizable Small
r4.large15250 MB21600 GB (EBS SSD)
r4.xlarge30500 MB41600 GB (EBS SSD)
r4.2xlarge61000 MB81600 GB (EBS SSD)
r5.large1600 MB21600 GB (EBS SSD)
r5.xlarge32000 MB41600 GB (EBS SSD)
r5.2xlarge64000 MB81600 GB (EBS SSD)
Use-Cases: Resizable instances allow rapid scaling (up and down) or processing capacity of the cluster using Instaclustr’s Dynamic Scaling capability. These instances can offer excellent value for money if you have predictable peaks and troughs in processing requirements.

While scaling-out, where more nodes are added to a cluster to increase processing capacity, takes days or even weeks, depending on the amount of data to re-distribute data and offer its full processing capacity, scaling up using resizable instances can increase processing capacity to its fullest in minutes. In general, for gradual and long-term increase in processing requirement, adding more nodes to a cluster is the recommended strategy, and, when you need higher processing capacity for a short duration, resizing the instances is the recommended strategy.


Microsoft Azure

Instance NameAzure Instance TypeMemoryCPU CoresStorage Capacity
Premium: TinyStandard _DS2_v27000 MB2256 GB (Premium SSD)
Use-Cases: Smallest available production node. Use this when getting started with minimal production traffic. When you need more capacity than a 3 node cluster, we recommend scaling up to DS12 or DS13 nodes rather than scaling out with more DS2 instances.
 
 
Premium: Small
Standard_DS12_v2
28000 MB
4
512 GB (Premium SSD)
Premium: Balanced1023 GB (Premium SSD)
Premium: Bulk Storage2046 GB (Premium SSD)
Premium: Extra largeStandard_DS13_v256000 MB82046 GB (Premium SSD)
Use-Cases: This family provides a range of different ratio of disk to processing capacity to allow you to select an instance that best matches the profile of your workload.

GCP

Instance NameGCP Instance TypeMemoryCPU CoresStorage Capacity
Tinyn1-standard-27500 MB2250 GB (Persistent Disk SSD
Use-cases: Smallest available production node. Use this when getting started with minimal production traffic. When you need more capacity than a 3 node cluster, we recommend scaling up to 4 core nodes rather than scaling out with more n1-standard-2 instances.
 
 
Small
n1-standard-4
15000 MB
4
400 GB (Persistent Disk SSD)
Balanced800 GB (Persistent Disk SSD)
Bulk1600 GB (Persistent Disk SSD)
High Memory: Small
n1-highmen-4
26000 MB
4
400 GB (Persistent Disk SSD)
HIgh Memory: Balanced800 GB (Persistent Disk SSD)
High Memory: Bulkn1-highmen-426000 MB41600 GB (Persistent Disk SSD)
Use-Cases: This family provides a range of different ratio of disk to processing capacity to allow you to select a instance that best matches the profile of your workload. The additional memory of highmem instances can provide improved price/performance for many use cases and is generally our recommended starting point.

Apache Kafka

AWS

Instance NameAWS Instance TypeMemoryCPU CoresStorage Capacity
EBS: tiny 500
r4.large
15250 MB
2
500 GB (EBS SSD)
EBS: tiny 800800 GB (EBS SSD)
EBS: standard 750
r4.xlarge
30500 MB
4
750 GB (EBS SSD)
EBS: standard 15001500 GB (EBS SSD)
EBS: standard 22502250 GB (EBS SSD)
EBS: high throughput 1500
r4.2xlarge
61000 MB
8
1500 GB (EBS SSD)
EBS: high throughput 30003000 GB (ESB SSD)
EBS: high throughput 45004500 GB (EBS SSD)
Use-Cases: This family provides a range of different ratio of disk to processing capacity to allow you to select an instance that best matches the profile of your workload. The additional memory of r5 instances can provide good price/performance for many use cases and are generally our recommended starting point for a production deployment.
 
 
Max throughput standardi3.xlarge30500 MB4884 GB (Local SSD)
Max throughput largei3.2xlarge61000 MB81769 GB (Local SSD)
Use-Cases: These instances come with local SSD and are suitable for application workload that require lower I/O latency and higher I/O throughput.

Microsoft Azure

Instance NameAzure Instance TypeMemoryCPU CoresStorage Capacity
Tiny 512
Standard_DS11_v2
14000 MB
2
512 GB (Premium SSD)
Tiny 768768 GB (Premium SSD)
Standard 768
Standard_DS12_v2
28000 MB
4
768 GB (Premium SSD)
Standard 15001500 GB (Premium SSD)
Standard 22502250 GB (Premium SSD)
High throughput 1500
Standard_DS13_v2
56000 MB
8
1500 GB (Premium SSD)
High throughput 30003000 GB (Premium SSD)
High throughput 45004500 GB (Premium SSD)
Use-Cases; This family provides a range of different ratio of disk to processing capacity to allow you to select an instance that best matches the profile of your workload. The additional memory of DS instances can provide good price/performance for many use cases and the Standard range is generally our recommended starting point for a production deployment.

Google Cloud

Instance NameGCP Instance TypeMemoryCPU CoresStorage Capacity
Tiny 400
n1-highmen-2
13000 MB
2
400 GB (Persistent Disk SSD)
Tiny 800800 GB (Persistent Disk SSD)
Standard 750
n1-highmen-4
26000 MB
4
750 GB (Persistent Disk SSD)
Standard 15001500 GB (Persistent Disk SSD)
Standard 22502250 GB (Persistent Disk SSD)
High throughput 1500
n1-highmen-8
52000 MB
8
1500 GB (Persistent Disk SSD)
High throughput 30003000 GB (Persistent Disk SSD)
High throughput 45004500 GB (Persistent Disk SSD)
Use-Cases: This family provides a range of different ratio of disk to processing capacity to allow you to select an instance that best matches the profile of your workload. The additional memory of n1-highmem instances can provide good price/performance for many use cases and the Standard range is generally our recommended starting point for a production deployment.

FREE TRIAL

Spin up a cluster in less
than 5 minutes.
(No credit card required)

Sign Up Now
Close

Site by Swell Design Group