Concat AG's annual symposium, on dsmISI / ISP+ / Dell PowerScale (OneFS) / VMware / Veeam provided customers with a forum to exchange information on products from our partners, Concat AG and General Storage on May 11, 2023.
The event focused on the combination of the Dell PowerScale (formerly Isilon) scale-out NAS series as central enterprise backup storage for the data protection products IBM Storage Protect, IBM Storage Protect Plus and Veeam. By using the General Storage dsmISI products, a high level of parallelization is achieved across all scale-out nodes through optimal load balancing of all data streams. In addition, the dsmISI product line offers direct integration of PowerScale shares as NFS datastores for VMware and as Oracle backup areas. This broad integration makes it easy to leverage the central scale-out concept even when using different data protection products. The common OneFS eliminates the need for sizing backup storage for individual applications and servers. Thus, a higher degree of utilization is achieved while simplifying administration and expandability. Once total storage exceeds a threshold, another scale-out node is simply added and the additional capacity and compute power is directly available to all connected servers. This completely eliminates the need to plan capacity and performance at the level of individual servers. The 6th Concat Symposium again offered the opportunity to learn about the latest developments, ask questions and benefit from the practical reports.
An overview of dsmISI products can be found at our partner, Concat AG.
Use Case: Complexity compensating use of technology.
The combination of Dell PowerScale and dsmISI is at the heart of the largest backup environment Empalis currently operates for customers. Here we back up data via Veeam and IBM Storage Protect using dsmISI, which gives us extremely high parallelism in the data stream.
The use of OneFS allows us to scale flexibly in data growth. It also eliminates the need to pre-provision storage to individual backup servers, leaving this capacity unused if the planned sizing is not suitable due to changes in requirements. This simplified planning allows us to react quickly and flexibly to unpredictable requirements, both in the Veeam, IBM Storage Protect and Oracle environments. Monitoring and planning are carried out centrally for the entire backup storage; there are no data silos.
High availability with General Storage Cluster Controller
We achieve high availability for IBM Storage Protect (formerly IBM Spectrum Protect/TSM) by using the General Storage Cluster Controller (GSCC). This product enables centralized management of the high availability of backup instances in a 2-node hardware cluster. Pivoting of individual or all instances is possible at the push of a button via the GUI. In the event of a node failure, failover runs automatically. As a special highlight, IBM Storage Protect is operated in a container, which is managed with Docker/Podman under RHEL (Red Hat Enterprise Linux). This type of deployment reduces update times of the IBM Storage Protect software by simply swapping the container images, so we are able to deploy new versions at shorter intervals to provide additional features and bug fixes.
Fast-clone technology reduces total memory requirements
The introduction of Fast Clone technology in Veeam v10 on Linux with XFS Reflink further reduces the overall storage footprint and moves Active/Synthetic Full Backup jobs to the repository server. The Reflink functionality is here integrated into dsmISI4v and still supports a high parallelism of mounts (on average 40), but files are stored on these DFS containers, which in turn are combined into an XFS file system. This is how we achieve deduplication via extent.
NAS backup via MAGS
NAS is backed up via MAGS, which also ensures us a very high data flow through parallelism. At times, 150 dsmc processes run simultaneously with tens of thousands of processes in the queue. This process allows us to successfully backup NAS shares with several billion objects in finite time.
Management of data volume and retention periods
Due to the amount of data and long retention periods, the data here is only written temporarily to Isilon/PowerScale. A migration of the data to tape already starts during the backup. Here, a strength of IBM Storage Protect is used, so that data can be moved completely transparently for the backup clients in the backend. Thus, the backup continues to run, new data lands on the backup storage with high performance and is displaced to the tape pools in the background without any time pressure.
Conclusion
The Concat Symposium again showed how important the detailed exchange with technical depth is. In particular, use cases and inquiries about products and applications that are not used to the same extent by customers, such as Empalis, which moves large amounts of data every day, conveyed comprehensive added value. It also became clear that experience gained in practical operation is always incorporated in order to become more proactive and optimize operations.
If you have any questions about dsmISI or Enterprise Architectures in general, please contact us.
