IBM Spectrum Scale Reviews

I am not certain what exactly was meant regarding policy-based data management and how it has helped streamline data handling in my organization.

Regarding the global namespace feature in IBM Spectrum Scale, I am not sure how to respond.

For integration capabilities with cloud platforms in hybrid cloud deployments, I have not personally worked on any integration with other vendor-based storage solutions. Migrations are much simpler, though we had to use a third-party tool called Komprise to migrate NAS data from existing infrastructure to GPFS or from GPFS to any cloud platform. Data handling is straightforward, but we need to rely on third-party tools. Unlike ONTAP which has VSM supporting direct migration from other storage devices to NetApp, with GPFS we must rely on third-party tools or Linux XCP, which is traditional and slower. For higher performance and high-speed transfer of data or migrations, Komprise is recommended.

IBM Spectrum Scale being software-based storage allows users to utilize any hardware supported or defined by IBM without dependency on physical hardware provided by IBM. This is one of the features I appreciate most because if you have existing infrastructure at your site but don't want to reinvest in hardware, you can purchase IBM Spectrum Scale software and deploy it as software-defined storage.

Users can comfortably access data from their host to the storage host without relying on new network setup when deploying GPFS as it uses the existing corporate network. However, high-speed ethernet network is required. By adding switches to the same network where client hosts reside and deploying GPFS storage array, any file system created on the storage cluster can grant export permissions determining which hosts can access specific file systems. These hosts can directly mount them as remote file systems. The process is simpler compared to defining IQN for providing access to client hosts. Creating a cluster for clients with virtual machines or physical Linux/Windows machines requires installing GPFS services on client nodes, which creates a client cluster. Access is defined on the storage cluster for the particular client cluster, allowing all hosts in that client cluster to mount file systems with granted permissions.

The system relies heavily on ethernet networking, and when issues occur, we must consult with the network team to investigate. While logs may indicate network errors, the situation differs from previous SAN setups where Fiber Channel was under our control. The storage team could directly identify and resolve Fiber Channel network issues. GPFS should consider enhanced integration with network monitoring to enable quick detection and resolution of network layer issues rather than depending extensively on the networking team.

From a functionality perspective, I cannot think of additional improvements or enhancements needed for the future.

Initially, stability was a concern as we encountered numerous issues and errors. However, after working with the system and tuning various parameters according to our application loads and requirements, it has become much more stable. The stability largely depends on providing accurate application details during initial deployment and properly setting parameters to optimize operations. This approach helps reduce daily alerts and issues.

While IBM Spectrum Scale is scalable, we have been deploying different clusters for redundancy purposes rather than scaling existing clusters. In my current organization, we follow a standard where setups must maintain uniform configurations including size and number of hosts. Instead of extending or resizing existing setups, we implement additional setups with identical configurations to maintain uniformity. Though scalability is possible, we have not explored that option with existing clusters directly.

IBM's software team provides excellent technical support and customer service. Their team is highly knowledgeable and provides detailed explanations rather than vague responses. When issues are reported, they respond with thorough explanations and solutions. Compared to other vendors, my experience has been exceptional.

I would rate IBM's technical support eight out of ten. They demonstrate commitment by joining calls even for minor issues, explaining situations thoroughly. As IBM Spectrum Scale is a newer product in the market, their helpful approach in explaining functionality and resolving issues has been particularly valuable.

Positive

I have worked with traditional storage solutions, but IBM Spectrum Scale was something new that I started working with four years ago. It has been an interesting experience, and I have made it my primary skill which has significant market demand.

While I did not personally make the decision to choose IBM, I am confident that my team evaluated various options before selecting this solution. IBM's new deployment option, ESS as an appliance, offers simplified implementation as it comes as a complete rack and stack solution that can be readily deployed in the data center.

For storage professionals working with IBM Spectrum Scale, the initial setup is not straightforward. Strong Linux expertise is required because traditional storage administration and deployments differ significantly from IBM Spectrum Scale implementation. These approaches represent distinct aspects of storage management.

IBM Spectrum Scale is definitely cost-effective. The investment provides 100% return on investment.

IBM Spectrum Scale functions as software-based storage, allowing users to utilize any hardware supported or defined by IBM without depending on physical hardware provided by IBM. If you have existing infrastructure at your site but want to avoid reinvesting in hardware, you can purchase IBM Spectrum Scale software and deploy it as software-defined storage, making it a cost-effective solution.

I have not personally tested or tried other products. I have worked with traditional storage solutions, but IBM Spectrum Scale was new to me four years ago. It has become my primary skill and has significant demand in the market.

IBM Spectrum Scale functions as a file system-based storage system, differing from traditional NAS or SAN storages. For organizations seeking cost-effective solutions with enhanced performance in file storage, this solution should be considered as an alternative to NetApp ONTAP or other market-standard NAS solutions.

The cluster-based deployment provides optimal availability and performance compared to traditional two-controller or four-controller configurations. In this scale-out architecture, both compute and storage can be expanded using basic disk configurations while achieving maximum performance through distributed reads and writes across physical drives within recovery groups.

The system utilizes GPFS native RAID instead of traditional RAID, distributing data across physical drives for improved resiliency compared to traditional storage solutions. Permission management and user access are streamlined, and IBM provides simple APIs that facilitate easy automation through REST APIs.

For data movement and migration, users can employ Linux-based XCP to copy data between source and destination, or utilize Komprise for faster, large-scale data handling.

I rate IBM Spectrum Scale seven out of ten as a complete product.

I use IBM Spectrum Scale to improve performance in data processing tasks. I have employed it in various projects, including a major project at Leading German Bank focused on the improvement of the FDW (Finance Data Warehouse). The primary goals were to reduce elapsed time and costs for regulatory reporting.

I find IBM Spectrum Scale to be an excellent product known for its fast parallel file system. It achieved the best results when integrated with IBM hardware. Even though it is complex, it provided significant performance advantages in large-scale data management. Its fault tolerance mechanisms and integration capabilities make it popular for extensive AI initiatives and data processing tasks for organizations like Daimler Benz and Bosch.

There can be improvements in fault tolerance and making erasure coding faster. 

Additionally, it would be beneficial to have example configurations or applications that assist in setup and configuration.

I have used it for nearly four years in various projects.

I find it stable, however, erasure coding can slow down the file system.

It scales well, however, it requires significant resources to avoid slowdowns from erasure coding.

I received excellent technical support. Customer service in my projects, particularly with the Leading German Bank project, was excellent.

Neutral

Previously, I evaluated and used Hadoop with Leading German Bank from 2016 to 2019, along with other products like Apache Spark. 

However, I found IBM Spectrum Scale gives better performance results.

The initial setup is complex, especially if trying to avoid erasure coding, as it requires more discs. Avoiding erasure coding can significantly increase costs.

IBM Spectrum Scale is very expensive with complex pricing models usually based on the amount of storage used or the number of servers. Using Excelero for mirroring storage helped reduce Spectrum Scale licensing costs by about half during the Leading German Bank project.

I evaluated Weka and Apache Spark, however, I found IBM Spectrum Scale to offer better performance for large-scale installations.

Although I used to be a big fan of Weka, due to recent limitations and lack of product improvements, I prefer to recommend IBM Spectrum Scale over others. 

I rate this product ten out of ten.

Positive

We are using GPFS Scale for parallel file access. It's primarily for file management.

We use GPFS Scale for parallel file access, the file management, and it's an essential part of our operations. It's the only feature we are using since we have not explored other options.

I wish there were some graphical user interface to access the GPFS file system creation and monitoring.

We have been working with Spectrum Scale for about ten years.

We have been using IBM's solution for a long time without encountering any troubles or problems. Stability is excellent, so I rate it a nine out of ten.

It's fully scalable. As it's a licensed product, I don't foresee any problem in scaling it.

In that regard, I can't give a very good comment since we never got the chance to escalate and get support from IBM.

Positive

Originally, we deployed this application with IBM systems. This was the recommendation, and we were satisfied, so we never looked for alternatives.

The setup was not that complex because we got vendor support. Certain changes are now performed by ourselves. For GPFS, it's not very complicated.

I don't think it's a high price compared to alternatives.

We never got a chance to explore any other solutions as we have been happy with this one.

For general use, I rate Spectrum Scale an eight out of ten. If there's a requirement for parallel access on IBM platforms, I recommend this solution.

The profile share is a valuable feature. We only use the parallel file system at my company.

I rate this solution a nine out of ten. I believe there is no graphic user interface, so they should include it.

We have been using this solution since 2014, and it is deployed on-premises.

It is stable and has not given us any trouble. We are a national payment institute, so we facilitate all the inter banks and some ATMs, so I am unsure how many people use the solution. We have about six people using IBM Spectrum Scale.

IBM completed the initial setup.

We are using the standard license. In addition to this standard license we purchased separately, we have IBM DB2 pureScale. I believe we got a bundled size because we purchased all the hardware.

We have had no issues with the solution, and I would recommend it.

Spectrum Scale is one of the most important products in our environment now. We are using it for monitoring all of our storage. We have about 60 users.

Integration with other vendors is not available. This feature would be very helpful.

We have been using IBM Spectrum Scale for four to five years.

it's stable. We had some issues before while deploying but it is good now.

IBM Spectrum Scale is scalable.

I would rate IBM Spectrum Scale at a six out of ten.

The setup was pretty complex. It takes several days to deploy the network advisor.

We used a consultant from IBM for our deployment.

We primarily use the solution for a parallel file system. I have one big file system for a thousand or so users.

The high performance of the solution is its most valuable aspect. If you compare it to other storage solutions, it's much better.

The pricing is pretty good.

The solution would benefit from being simplified.

The scalability could be a bit better. It needs to be more flexible when it comes to expanding.

It would be helpful if there was a graphical user interface that could walk you through the deployment process. The instructions surrounding setup aren't the best. They need to be more step-by-step.

We've been working with the solution for a couple of years at this point.

The solution is pretty scalable. I'd rate it a ten out of ten for scaling. 

I've contacted technical support in the past. I've had a couple of cases that I've had to pass to them to assist with, and they were great. They are knowledgable and responsive.

They are also really helpful with a variety of things. I remember once I had to download some software to demo, IBM technical support gave me a demo without charging me. It was a very nice gesture.

The initial setup is pretty complex. It's not too straightforward. 

To deploy it, it takes about one week.

I'd rate the deployment process at a six out of ten. 

I didn't need a consultant or reseller to assist in the initial setup. I handled the implementation myself.

The pricing isn't too expensive. It's reasonable.

We're partners with IBM.

We use both on-premises and hybrid deployment models.

I'd rate the solution overall at eight out of ten.

I don't know too many other products, however, I'd recommend either IBM or NetApp to others to use if they need a similar solution for their company.

I am a pre-sales engineer in an IT company and we provide a variety of solutions to our clients. IBM Spectrum Scale GPFS is one of the products that we resell and I have experience with.

We provide IBM Spectrum scale as a high-availability storage solution. One of our customers is in the business of making and transforming videos and they use different platforms such as Linux, AIX, and Windows. Using this solution allows access to all of the files on all of these platforms from a single place.

The most valuable feature is the ability to share files across different platforms. It is platform agnostic and not attached to IBM, AIX, Windows, or Linux.

The pricing and licensing model for this solution are complex and it is sometimes difficult to explain it to customers.

It is not always easy to sell this product because certain aspects seem complicated for the customer.

I have between ten and fifteen years of experience with this product.

This is a stable solution.

Scalability is very important and it is one of the reasons that we chose to sell and market this product. Our clients vary in range from small companies to big banks that have a lot of servers, so this solution works regardless of the company size.

I do not speak with technical support, although I am close to the installation process and I do not recall any problems that required contacting support for help.

I have not personally installed this solution but it is my understanding that it is pretty simple. There have been no problems that I have heard of.

The deployment can take a few days. It depends on different phases because you have to meet with the customer and it may influence what you are going to do. Installing it is very quick and then you move into the testing phase, which takes a little longer.

We provide our customers with assistance during the deployment by way of knowledge transfer. This can last a few days. If they ask questions then we give them support.

The licensing model is complex and depends on factors such as the number of processing cores and the amount of storage.

This is a very powerful product. It can be optimized for performance, although our clients do not use it with this aspect in mind. Over the past five or ten years, IBM has developed this solution and it has improved a lot. With the latest release, I think that this is a very complete product.

I would rate this solution a seven out of ten.

The primary use case of this solution is for virtualization, data restore for VMware, NFS exports for a backup share, and in the future, we are thinking about using this to update based storage for S3 buckets.

At the moment, we mainly use it for NFS topics.

GPFS monitoring is the best feature. The performance is very good and so is the scalability. 

The NFS CL solution is impressive because if you have a power loss for any reason from the consumer host, the system switches seamlessly to the next available node in this cluster. This happens without any interruption in the data stream, reading or writing. 

When the system is coming back from the maintenance or the reboot cycle, the system switches automatically back to the initial host. This is very good and very friendly. I have never seen this before when comparing to other solutions such as NetApp, and Dell EMC Unity. NetApp is also very good for NFS solutions.

The biggest problem is that it is not able to provide block storage. It can consume block storage, you can put it on from the storage and present the span over the fiber channel or iSCSI. The problem is that they have not been able to provide block storage consumable to the customer host. This is only a side-based solution. We can install the GPFS into the customer environment but I don't think that is the best solution.

The initial setup has room for improvement, it is not straightforward.

Spectrum scale is very good from the performance perspective, but from the application perspective, you have to take care if you want to use a database as a service that you can consume. For example, a database, or being able to read and write it, doesn't have a deployment solution. It is missing.

I have been using this solution for two years.

We started with version 502 and now 503.1.3.2 the lastest and the greatest version.

This solution is stable, it's very good.

This solution is scalable it's very good. With this solution running, you have solved some cosmetic issues. The system is very stable.

The biggest issue we have faced is the complete interruption of the network. We have tested a complete site loss with the initial loss of the complete data center. We tested this and the system was always up. This is very good.

We have 10,000 clients, so there may be 100,000 to 200,000 users. I am not sure.

The technical support is good. 

At this moment we have a problem with technical support but it is not related to Spectrum Scale. It's with another product called Tivoli Suite, Spectrum Control Site.

Currently, we are also using Nutanix, Vison, ScaleIO, and Huawei as an STS solution. The problem is that we want to use one solution in the future and not five.

I have to compare all of these and decide which one is the best for the future.

The initial setup was complex. We had to prepare a lot of the requirements to install the Spectrum Scale. It is not straightforward. We had to ensure that the foundation for the install and that the installation process was proper.

On average for large clusters, the deployment can take one week.

We did not use the help of an integrator, vendor, or reseller. I implemented this solution.

The licensing is based on the number of terabytes. 

We have an enterprise license agreement with IBM and it doesn't reflect the moment price list.

I think that it's a good price but it is quite expensive.

We are using this solution in our Opcos and the challenge in my role as an architect is staying informed of what solutions are to be used in the future. What is the best solution to carry that will cover all of the requests that we have and for the multiple cloud solutions regarding STS soft topic?

I am currently making comparisons for this and I need support for this as a strategy for the next five years. I am the lead architect for this solution in our company.

My advice would be to use a consultant or a technical person to implement this solution. It's a better way.

I would rate this solution an eight out of ten.

We use it a lot for sharing file systems between several Red Hat and AIX servers. It is mostly for different lines of business. It works pretty well. We have been using it for almost a decade now.

It is actually one of the better products of IBM. We have seen applications come and go, but Spectrum Scale is one of those things where we have had more customers move into the product than other vendors' solutions.

It makes our file system sharing a lot easier, even across different continents. We have had file systems shared across different continents with no performance degradation. In this sense, it has helped us. 

The ease of use and the stability of the product. It has been pretty reliable throughout the years. As far as capacity is concerned, it can handle most heavy loads, unlike NFS or even NAS. In this particular case, it is seamless. That is the biggest thing.

Maybe it needs integration with HA.

Stability is one of the main reasons that we are using the product. So far, it has been a very reliable software.

We have had it across 15 to 20 Clients. Especially now that we are using it across Red Hat, the Clients do not even know that it is sitting on AIX servers. As far as scalability is concerned, it is only beginning to show its strength in the past few years. We are using it on Red Hat, but still using NSD servers as the main source of storage. 

As far as scalability, I would support this product more than anything that I have tried before, such as NFS or even HACMP. This product is more robust in my point of view.

We still use NAS and NFS (mainly NFS), which we used prior to Spectrum Scale.

We added Spectrum Scale because of its scalability. There are just large amounts of data, especially with Ab Initio, NFS would not cut it.

It is not that complex for me since I have been working with it for years. 

One thing that is new for me is the implementation across Red Hat Clients. This portion was not as straightforward as I would like it to be. Now that we have it documented, it is about 10 steps, though more steps than I would do if I were to mount it on AIX. They should probably simply the Red Hat implementation portion.

There is nothing else that even compares to Spectrum Scale. We have had customers who switched to AIX just because of it.

It is doing more than it is expected to do as far as file sharing.