By Roger Freixa, Director Managed Services CPSBU, VMware
During VMworld 2018 U.S., VMware and AWS announced Amazon Relational Database Service (Amazon RDS) on VMware. See the announcement on our Cloud Community blog here.
Amazon RDS on VMware is a service that makes it easy for customers to set up, operate, and scale databases in VMware-based software-defined data centers and hybrid environments. RDS on VMware automates database provisioning, operating system and database patching, backup, point-in-time restore, compute scaling, instance health monitoring and failover.
Amazon RDS on VMware supports Microsoft SQL Server, PostgreSQL and MySQL. Other databases will come after GA, so stayed tuned for those announcements.
Amazon RDS on VMware takes advantage of durability and resiliency features of VMware vSphere 6.5 and 6.7. The service has been certified to use most of the resiliency, durability and high availability features available on vSphere (HA, DRS and vMotion). For more information, please visit https://docs.aws.amazon.com/AmazonRDS/latest/RDSonVMwareUserGuide/getting-started-with-rds-on-vmware.pre-onboarding.html.
The use case for RDS on VMware is to run on customers’ on-premises VMware environments. For that, each vSphere cluster can be mapped to a corresponding Amazon RDS on VMware custom Availability Zone. At this point, the solution only supports a 1 to 1 mapping, however, a customer can have as many vSphere Clusters onboarded as needed.
vSphere Requirements and Architecture
VMware and AWS have provided detailed documentation on requirements for running the service. The most important criteria is that customers must have vSphere Enterprise Plus Edition or higher and be running VMware vCenter Server and ESXi versions 6.5/6.7. Additionally, all ESXi servers on the cluster should be connected to a shared Datastore. An onboarded vSphere Cluster must support the following network architecture, with access to 4 different networks:
| Networks | Description |
| Internet
Network (1) |
Customer managed. This can be an existing network with outbound internet access. Main goal of this network is to establish the site-to-site VPN that RDS on VMware creates and maintains automatically. Requires DHCP services, provided by the customer |
| Application Network (2) | Customer managed. Existing network where the database instances will be deployed. Each database will also have an interface to the Cluster Control network. This network is where Database instances will offer the SQL access for client applications. Requires DHCP service, provided by the customer |
| Application Network (3) | AWS managed. Guest network where the database instances will be deployed. Each database will be used only to bind RDS on VMware components. This network will run a DHCP Server provided by RDS on VMware (once the Edge Router appliance is deployed). RDS on VMware assigns IP addresses in pre-defined 54.239.236.0/22 range of public IP addresses that are not internet-routable. Amazon RDS on VMware deploys a DHCP Service and a DNS resolver on this network
If the customer |
| ESXi Management Network (4) | Customer managed. Such network is present in most of the vSphere installations where ESXi Hosts and vCenter are bind on a relatively isolated network which hosts only ESXi related traffic. |
Another important network requirement is to configure your local DNS servers to forward requests for *.rdsonvmware.rds.amazonaws.com to one of the IP addresses on the RDS Edge Router VM. Specifically, the IP address of either the Management Network (4) or the Internet network (1). Amazon RDS on VMware installs a DNS resolver on the Custom Control Network.
One of the important parts of the product is to allow client applications to connect transparently to Amazon RDS on VMware managed DB instances from any network connected to the corporate network where the vSphere Cluster is running.
How to onboard the solution?
In order to use the service, a vSphere cluster must be onboarded to become an Amazon Web Services (AWS) custom Availability Zone. For this to happen, you must Sign into RDSthe AWS Management Console, select US East (N Virginia) Region and choose Custom AZs on the navigation panel.
While waiting for the download to complete, we can configure the new custom Availability Zone information which is the logical construct of RDS on VMware that maps to the vSphere Cluster that is being onboarded.
Once the download is completed, we will deploy the RDS on VMware Installer into the vSphere Cluster that we want to onboard, following the screens of a very simple installer. Be careful about selecting the proper networks at deployment time: there are four network interfaces, one for each of the networks described previously..
To start the installer, look at vCenter console for any of the IPs assigned to the installer and access to such IPs via browser, you should see the first page of the installer. Be careful about selecting the proper networks at deployment time: there are 4 network interfaces, one for each of the networks described previously.
Once the installation has successfully completed, the RDS Console will show that the Custom AZAZs status is now “Active” (in green). The corresponding vSphere Cluster will show the following objects deployed.
At this point, vSphere Cluster can be used as a target to host RDS on VMware Database instances that have been created from the RDS Console, CLI and RDS API (SDK). In order to create Database instances, you must select “On-Premises” and select the custom Availability Zone that has been created, like illustrated on the following image.
Each Database created from RDS Console (or via API or CLI) will have the corresponding DB instance VM showing up on the vCenter that is mapped as a custom Availability Zone. See the images below:
Once the database has been created, we can point which is the corresponding Virtual Machine on vSphere that hosts the RDS Database instance.
On the Amazon RDS console, select the Database instances and go to the Configuration tab, and you will see a similar screen:
You can see the highlighted field “Resource Id” that has a reference number. You can map this value to the corresponding vSphere VM deployed on the cluster: all the database VMs managed by RDS on VMware have a prefix “rds-VM” followed by the reference number.
See the corresponding VM on the vSphere Cluster
At this point, we have seen how a DB Admin (accessing primarily the RDS Console) and a vSphere Admin (accessing primarily the vSphere vCenter console) can work together to improve and diagnose issues on specific RDS Database instances deployed on vSphere.
The post How Amazon RDS on VMware Works appeared first on VMware vSphere Blog.