InfoScale™ 9.0 Solutions in Cloud Environments
- Overview and preparation
- Overview of InfoScale solutions in cloud environments
- InfoScale agents for monitoring resources in cloud environments
- InfoScale FSS feature for storage sharing in cloud environments
- InfoScale non-FSS feature for storage sharing in cloud environments
- About SmartIO in AWS environments
- Preparing for InfoScale installations in cloud environments
- Installing the AWS CLI package
- VPC security groups example
- Configurations for Amazon Web Services - Linux
- Configurations for Amazon Web Services - Windows
- Replication configurations in AWS - Windows
- HA and DR configurations in AWS - Windows
- EBS Multi-Attach feature support with InfoScale Enterprise in AWS cloud
- InfoScale service group configuration wizards support for EBS Multi-Attach
- Failover within a subnet of an AWS AZ using virtual private IP - Windows
- Failover across AWS subnets using overlay IP - Windows
- Public access to InfoScale cluster nodes in AWS using Elastic IP - Windows
- DR from on-premises to AWS and across AWS regions or VPCs - Windows
- DR from on-premises to AWS - Windows
- Configurations for Microsoft Azure - Linux
- Configurations for Microsoft Azure - Windows
- Replication configurations in Azure - Windows
- HA and DR configurations in Azure - Windows
- Shared disk support in Azure cloud and InfoScale service group configuration using wizards
- Failover within an Azure subnet using private IP - Windows
- Failover across Azure subnets using overlay IP - Windows
- Public access to cluster nodes in Azure using public IP - Windows
- DR from on-premises to Azure and across Azure regions or VNets - Windows
- Configurations for Google Cloud Platform- Linux
- Configurations for Google Cloud Platform - Windows
- Replication to and across cloud environments
- Migrating files to the cloud using Cloud Connectors
- About cloud connectors
- About InfoScale support for cloud connectors
- How InfoScale migrates data using cloud connectors
- Limitations for file-level tiering
- About operations with Amazon Glacier
- Migrating data from on-premise to cloud storage
- Reclaiming object storage space
- Removing a cloud volume
- Examining in-cloud storage usage
- Sample policy file
- Replication support with cloud tiering
- Configuration for Load Balancer for AWS and Azure - Linux
- Troubleshooting issues in cloud deployments
InfoScale FSS feature for storage sharing in cloud environments
InfoScale supports flexible storage sharing (FSS) in cloud environments for a cluster that is located within the same region. The nodes in the cluster may be located within the same zone or across zones (Availability Zone in case of AWS and user-defined site in case of Azure). FSS leverages cloud block storage to provide shared storage capability.
Storage devices that are under Volume Manager (VxVM) control are prefixed with the private IP address of the node. You can override the default behavior with the vxdctl set hostprefix command. For details, see the Storage Foundation Cluster File System High Availability Administrator's Guide - Linux.
VxVM provides the ability to grow the existing storage by growing the LUN, which is termed as Dynamic LUN Expansion (DLE). You can leverage DLE to grow the storage in cloud environments. To use the DLE feature, you invoke the vxdisk resize command with the length option from the master node in an FSS configuration. Alternatively, you can use the Management Server console of InfoScale Operations Manager to resize a disk, which internally invokes this command. The command intelligently detects the remote disks and gets the required protocol executed to complete the LUN expansion; you don't need to specify any additional options. No explicit master switching is required. For details on growing the existing storage by growing the LUN, refer to the Storage Foundation 9.0 Administrator's Guide - Linux.
Note:
Do not resize an LVM disk or any other disk on which the OS is installed.
In cloud environments, FSS in campus cluster configurations can be used as a disaster recovery mechanism, across data centers within a single region. For example, in AWS, nodes within an AZ can be configured as one of the campus cluster site, while the nodes in another AZ can be configured as the second site. For details, see the Acrtera InfoScale Disaster Recovery Implementation Guide - Linux.
Note:
(Azure only) By default, in addition to the storage disks that you have attached, every virtual machine that is provisioned contains a temporary resource disk. Do not use the temporary resource disk as a data disk. A temporary resource disk is an ephemeral storage that must not be used for persistent data. The disk may change after a machine is redeployed or restarted, and the data is lost.
See About identifying a temporary resource disk - Linux.
For details on how Azure uses a temporary disk, see the Microsoft Azure documentation.
Note:
(Azure only) In case your configuration uses Premium SSD v2 disks, remember that they are shared across all the VMs in a zone, by default. Such disks can be unintentionally attached to non-InfoScale-cluster VMs, without any error or warning. To restrict the disk access to InfoScale cluster nodes and thus prevent unauthorized use, set the Disk.MaxShares property of the Azure disk to 1.
Note:
(GCP only) When VCS is stopped and started on VM instances, or after a node restarts, the import and recovery operations on FSS diskgroups may take longer than expected. The master cannot import a diskgroup until all the nodes have joined the cluster. Some nodes may join their cluster with some delay. In that case, a diskgroup import operation takes longer than expected to succeed. Even if the master initially fails to import the diskgroup due to such a delay, the operation completes successfully later on retry.
In cloud environments, FSS is supported with LLT over reliable UDP configurations only. Acrtera recommends that you configure LLT over UDP for both, FSS and non-FSS clusters in the cloud.
Cloud-based networks are relatively slow and have high latency as compared to physical networks.
To achieve better LLT performance in such high-latency cloud networks, set the following tunable values before you start LLT or the LLT services:
set-flow window:10
set-flow highwater:10000
set-flow lowwater:8000
set-flow rporthighwater:10000
set-flow rportlowwater:8000
set-flow ackval:5
set-flow linkburst:32
For details on the usage of these tunables, refer to the Cluster Server Administrator's Guide.