What is our primary use case?
When we started using Dell PowerScale (Isilon), our primary use case was to have a general purpose file server, which can integrate with all of our domains or tenants, and in which we can provide SMB and NFS services because S3 was not part of it seven years back. The next case was to have data access between multiple tenants, so you authenticate one tenant and then write your data, and then you can authenticate from another tenant and get your data or read your data and do vice-versa.
What is most valuable?
Dell PowerScale (Isilon) includes its replication features; when we decommissioned our old Isilon cluster, we migrated all of the data to a new cluster, and it was pretty seamless. All you have to do is create the certificates and make those certificates available to the target cluster, then initiate a few jobs. The inter-cluster replication feature, called SyncIQ, allows you to set up all of the jobs and move your data entirely, either timely or all at once. Another suitable feature is the ability to back up your directories; you can take snapshots for all of your directories, and in fact, I believe you can take around a thousand snapshots for a single directory.
The inter-replication, geo-replication is one feature. Dell provides a variety class of hardware suitable for your workloads, meaning if your data is just lab and not critical, you don't have to spend much on the hardware; conversely, if your data is mission-critical, you can buy a high-performance node from Dell. Additionally, you can span Dell PowerScale (Isilon) cluster to a large number of nodes, and code upgrades are pretty smooth because all protocols will be upgraded under a single piece of code. Since it provides a single file system that spans across the nodes, you don't have to worry about that directory structure being separately created on individual nodes, making it a fitting solution for a data lake solution because you can access the same data from multiple protocols.
What needs improvement?
Dell PowerScale (Isilon) certainly has room for improvement; since it does not provide SAN at the moment, I would look for Dell to create a complete unified solution just like NetApp. The missing protocol on Isilon is block storage accessible over NVMe, as Fibre Channel protocol is not the future anymore. Many users exploring options like NVMe would appreciate this integration. I'm not certain about backups, but many backup vendors are now offering D-NAS backups for taking snapshots of your file shares. Isilon currently does NDMP but could benefit from advancements in backup solutions.
For how long have I used the solution?
I have been using Dell PowerScale (Isilon) for seven years.
What do I think about the stability of the solution?
In terms of stability and reliability, I would rate this a ten. The way we write our data on Dell PowerScale (Isilon) utilizes a protection plan that enables maintaining two to three blocks of your data across all nodes. Even if a node goes down, other nodes can respond and provide data back to the user without issues, and while there can be hardware failures, they are rare.
I would rate stability a ten because it is a very stable product.
What do I think about the scalability of the solution?
Dell PowerScale (Isilon) is scalable; our first cluster had twelve nodes, and since we can mix and match our nodes, we currently have four clusters with hybrid nodes, which fits our performance and capacity needs.
How are customer service and support?
I would rate the technical support at a seven. It was great earlier, but now, when we encounter issues, level one support often needs level two assistance. This delays the resolution process, which has led to perceived degradation in support quality, compared to the amazing support I received seven years back.
How would you rate customer service and support?
Which solution did I use previously and why did I switch?
I set up Dell PowerScale (Isilon) for my organization from ground level up. We had a different tool at that time, which is called Arcserve, Arcserve RHA, and we decommissioned that tool and migrated all of its functionality over to Isilon.
How was the initial setup?
My deployment experience with Dell PowerScale (Isilon) is not complex. When we first set up the cluster, we received help from professional services, who provided all environmental details, infrastructure, network tenants, user base, IP ranges, subnets, and more. Once we filled out a sheet with that information, they pre-configured the box for us and assisted via Zoom for setting everything up based on our needs. The setup process is quite smooth, and as we integrated more network tenants, the procedure remained straightforward.
What about the implementation team?
The deployment time can vary; you have multiple dependencies with different teams, such as needing IP addresses or CIDR ranges from peer network teams. However, if your team has available resources and they can assist quickly, such as within a day or a few hours, it won't take much time to deploy.
What was our ROI?
The return on investment with Dell PowerScale (Isilon) has been significant, but it depends on your use cases. For instance, if you create a share on AWS, AWS won't charge you until you pull your data out, and if your lab requirements are high, many companies prefer creating their lab environment on-premises instead of on AWS or other clouds, which is far cheaper. A similar environment on-premises can yield extraordinary savings, potentially more than fifty percent, because a single procurement of Isilon gives you three to four years of support and maintenance.
What's my experience with pricing, setup cost, and licensing?
Regarding pricing, Isilon can be cost-efficient; if you are using more capacity nodes without needing more performance, it won't affect your budget much. However, multiple factors play a role in deciding the cost; are you going to be using replication or encryption? Cheaper nodes will cost less, while higher performance nodes will be pricier. One more point to add is that if Isilon were to become software-defined, it would be a good option, because currently it is only hardware-based; you would buy hardware, set those up in your data center, and it is done. If Isilon were to be software-defined, users could create their own hardware and install Isilon's software on low-cost servers, which is not currently available.
Which other solutions did I evaluate?
When comparing Dell PowerScale (Isilon) with what is on the market, I can say the best competitor out there for Isilon is NetApp. NetApp does not provide you a spanned single file system architecture, so the file system will be only dedicated to a NetApp cluster. But the functionality will be the same, and people usually go for NetApp because it is a unified system; you can do multi-protocol, all sorts of storage protocols on NetApp. It provides you SAN, it provides you NAS, and it now provides you S3. However, it is a bit complex to manage—not very complex, but it is more complex than Isilon. Next is the upgrades; the upgrades are not pretty smooth for NetApp, and if anything goes bad during an upgrade, all your protocols and everything, all your systems will be affected. I am not sure about the costing feature because we evaluated Isilon and found it a perfect match. Plus, since you can map a single directory to multiple network tenants, you cannot do that on NetApp; you have to internally migrate the data from one directory to another that is respectively mapped to different network tenants. For example, if I have a network domain which is lab and I have another network tenant which is production, you can integrate a single Isilon hardware piece to both the domains, lab as well as prod, and you can easily restrict access to users in both network tenants without having to replicate the data.
What other advice do I have?
Automation plays a role in our AI initiatives, as there are APIs available for Isilon. We have streamlined our footprint when it comes to the data center, so we do not entertain many requests on a regular day. Automation is available, and if we want to expand our footprint later, we will probably go with the automation piece, but at this point in time, we don't get many requests, so we don't want to spend more on creating automation rather than having people available for using it. However, automation is available, and you can also use your workloads on Grafana, allowing custom creation of your performance metrics.
To measure the success with Dell PowerScale (Isilon), we look to the performance metrics and benchmarks we have established. For example, we previously had F-series nodes in our last cluster, which are high-performance nodes, meaning you can actually move your data from one node to another if the data is least used. Additionally, you can integrate Isilon with any other third-party or Dell product, creating a storage pool for dumping your backups. Performance-wise, our primary goal was general-purpose file shares for SMB and NFS, so we aim to ensure that users do not face any sort of latency. Latency varies; if the shares are mission-critical, they require fast access, while lab environments can accept some latency, but we do have certain response time numbers to consider. For example, users should not experience timeouts when accessing SMB shares or NFS shares, and the metrics for latency should be under a defined threshold, not beyond ten milliseconds or potentially fifteen or thirty milliseconds, depending on the workload.
Dell PowerScale (Isilon) caters to all businesses, whether small, medium, or enterprise, because of the variety of nodes available.
I would recommend doing a proof of concept before implementing Dell PowerScale (Isilon), especially if an organization is migrating from an existing product. It is crucial to match the performance you are getting now with what Isilon is offering, considering your use cases. For example, we had to securely migrate PCI data to different network tenants, and Isilon allowed us to map directories to both lab and production, making it suitable for our needs. I give this product an overall review rating of ten.
Which deployment model are you using for this solution?
On-premises