Home > Workload Solutions > SAP > Guides > VMware Virtualized SAP HANA with Dell Storage > SAP HANA database persistence
SAP HANA persists the in-memory database to the /hana/data file system at regular intervals and writes the transaction log to the /hana/log file system. These two file systems are the SAP HANA database persistence. The storage performance requirements for SAP HANA TDI certification apply to the SAP HANA persistence for production systems only. Dedicate the underlying storage device (RDM or a VMware VMFS datastore) to the persistence and then configure it using the best practices for SAP HANA on that storage array. For more information, see the validated design guides for the relevant SAP TDI-certified Dell EMC storage system at the Dell Technologies info hub.
In production environments, a virtual SAP HANA node requires the same storage performance and bandwidth as a physical node. The configuration of the LUNs and the underlying pools or RAID groups must follow the same rules. Because the additional virtualization layer affects the bandwidth and latency, the scalability of a particular Dell EMC array for virtualized SAP HANA nodes might be lower than the number of SAP HANA nodes supported in a physical environment.
For nonproduction virtualized SAP HANA environments, the storage performance does not have to meet the SAP KPIs. This means you have more flexibility when you configure the virtual environment: RAID protection, disk types, pools, RAID groups, and so on. Dell EMC recommends monitoring the storage performance and making adjustments if the performance does not meet your requirements.
When creating LUNs for virtualized SAP HANA deployments in production environments, follow the storage configuration best practices for physical SAP HANA environments corresponding to the SAP TDI-certified Dell EMC storage array. Look for the relevant validated design guide at the Dell Technologies Solutions Info Hub for SAP.
When Dell EMC block storage devices (LUNs) are connected to vSphere ESXi hosts, the SAP HANA virtual machines can use them for the database persistence, either as datastores formatted using a Virtual Machine File System (VMFS) cluster or as raw device mappings (RDMs). VMFS datastores are preferred because they are easier to manage and provide more data growth flexibility.
The use of VMware Virtual Volumes (vvols) is outside the scope of this guide.
Creating a VMFS datastore on a block storage LUN is the preferred configuration option for SAP HANA virtual machines. Creating dedicated and isolated datastores for SAP HANA data and log file systems is a VMware best practice, which means the datastores must be created on different LUNs in the storage array. Create all data devices (VMDKs) of an SAP HANA system on one datastore, and all log devices on a second datastore. When running multiple SAP HANA databases in a vSphere cluster, you can achieve further workload isolation with dedicated data and log datastores for each SAP HANA database system ID SID.
Each SAP HANA worker virtual machine has a data and a log file system residing on VMDK files created on the corresponding VMFS datastore
The following figure shows an example of an SAP HANA three-node scale-out deployment on a vSphere cluster using VMFS datastores:
RDMs are an alternative option for the SAP HANA database persistence. With RDMs, every data and logfile system resides on a dedicated LUN in the storage array.
RDMs enable you to configure the LUNs for virtualized SAP HANA nodes the same way as for physical environments. They also enable you to use the LUNs for physical SAP HANA installations when the nodes are migrated to and from a virtualized environment. For larger SAP HANA deployments, consider VMFS datastores instead because the high number of LUNs might become difficult to manage.
Figure 6 shows an example of an SAP HANA 3+1 scale-out deployment on RDMs:
This section provides guidelines for correctly sizing the LUNs used as RDMs or the storage LUNs for the vSphere VMFS datastores. For information about capacity sizing for SAP HANA devices, see the SAP HANA Storage Requirements white paper in SAP note 1900823: SAP HANA Storage Connector API. The storage requirements have been updated several times, so ensure that you refer to the most recent version of the guide when sizing a new SAP HANA installation.
The examples in this implementation guide use the following capacity sizing:
The LUNs for the VMFS datastores contain the SAP HANA persistence. When sizing them, provide the following information:
Figure 5 shows a three-node SAP HANA scale-out cluster. Assuming each node has 128 GB RAM and holds approximately 50 GB of data, the capacity calculation is:
All VMFS datastores must retain a level of free space to ensure that the virtual machines function correctly. Dell EMC recommends adding 25 percent free space to the previous calculation. The size of the LUNs should be:
These are the minimum capacity requirements for the SAP HANA persistence. They do not take account of future growth. With virtualized SAP HANA, you can increase the RAM capacity of the SAP HANA nodes, requiring larger file systems for the database persistence and therefore more capacity in the VMFS datastores. It is more efficient to create larger LUNs at the beginning to prepare for future growth. It is also better to avoid adding another LUN to a VMFS datastore because performance does not scale when LUNs are concatenated.
Unlike physical environments, the underlying disk configuration of virtual environments--RAID groups, thin pools, and so on--does not change. You must still adjust the size and configuration of the LUNs to meet the capacity requirements for datastores.
When using RDMs for the SAP HANA persistence, size the storage LUNs in accordance with the SAP minimum requirements. For more information, see the SAP HANA Storage Requirements white paper in SAP note 1900823: SAP HANA Storage Connector API.
For the example shown in Figure 6, we created the following LUNs: