IBM reorients to offset historic storage hardware decline


In this storage vendor profile we look at IBM, which has a long history and in IT much more widely than just storage. It’s still a big player in servers, services and research and development (R&D).

If anything characterises the company now, it is the ongoing strategic change after a period of declining revenues. That re-orientation takes the form of a heavy bet on the cloud, containerisation and as-a-service modes of purchase.

In this article, we look at IBM, its origins, its key storage products and its approach to the cloud, containers and consumption models of storage purchasing.

Where did IBM come from? 

IBM is truly a giant – among corporations in general and not just IT, let alone storage.

Originally founded in 1911 as the Computing-Tabulating-Recording Company, it changed its name to the much snappier International Business Machines in 1924 and became a long-time leader in punch card information recording systems, then electric typewriters, calculators and other office machinery.

From the 1960s onward, it became synonymous with the roots of modern IT and was a leader in mainframe and later personal computing. It lost ground in the PC market to emerging players, got out of consumer products and focused on enterprise computing products and services, as well as research, where it holds 150,000 patents.

Perhaps the most far-reaching recent shift for IBM has been its acquisition of Red Hat for $34bn in 2019. That brought the Red Hat portfolio of Linux and cloud- and container-focused products – Ceph, OpenStack – to IBM and showed the company had put a sizeable bet on a hybrid/multicloud strategy.

How does IBM rank against other storage players?

By the second quarter of 2021, IDC ranked IBM joint fifth among the big six storage array makers, with market share (4.7%) in a similar ball park to Hitachi (4.9%) and Pure Storage (4.1%) for “external OEM storage systems”. Dell, HPE, NetApp and Huawei occupied the top four places in that ranking. IBM is in a state of relative long-term decline against other companies, including in tech.

In 1980, the Fortune 500 had IBM as the 8th largest US company, with no other tech company in the top 30. It ranked 65th in the 2023 Fortune 500, with many tech companies ahead of it, including Amazon and Apple.

IBM revenue has been on a downward trend over the past two decades, with averages of around $90bn from the turn of the millennium to around 2007 and reaching $106bn in 2011. It has declined since then.

However, it is still a giant. In 2022, IBM turned over $60.5bn in revenue and had around 300,000 employees in 170 countries. Profits for year ending 2022 were about $1.6bn with a profit rate of 2.7%.

IBM storage hardware revenues have mirrored that decline, with sales near $3bn in 2010 but around half that by 2021, but that figure does not include software sales, nor storage-related cloud figures, so it’s possible these have taken up slack from decreased hardware sales.

In IBM’s first-quarter 2022 results, revenue had grown by 8% after almost a decade of negative reports. Red Hat was a leading contributor with 18% growth, while the hybrid cloud business grew 14%.

What are IBM’s key storage products?

A big recent change has seen IBM drop its Spectrum brand, at least in part. So many of what were Spectrum this, that and the other have had added “storage” as part of their name.

IBM Storage FlashSystem arrays come in the 5000, 5200, 7300 and 9500 editions, that are differentiated by use of HDD, 2.5” SSD flash, NVMe flash and combinations thereof. Capacities also increase, ranging from about 0.5PB in the 5000 series to 4.5PB in the 9500, as does performance of controller CPU.

All FlashSystem arrays are hybrid cloud-capable, with connectivity to public clouds for data tiering, migration, replication and snapshots possible. That functionality comes as part of IBM’s Spectrum Virtualize, which all these products include.

IBM’s Storage DS8K family of flash storage is targeted at customers with IBM zSystem mainframe systems (via FICON) and Power servers (via Fibre Channel).

The two model sub-lines are the entry and midrange DS8910F 993 and 994 of 1.5PB and 2.9PB maximum capacity, plus the DS8950F 996 and 998 with up to 6PB capacity. All offer minimum access time of 80µs, or 18µs with the mainframe-optimised zHyperLink connection.

IBM claims seven nines availability for DS8K arrays. They come with full data encryption, immutable copies – Safeguarded Copy – to protect against ransomware, automated tiering and disaster recovery functionality. Other features include hybrid cloud connectivity, a number of deep integrations with IBM server and mainfame operating systems, and container storage via CSI.

IBM Storage Scale is the company’s scale-out storage for file and object. It is based on Storage Scale 3500 hardware appliance nodes that can scale from a few tens of TB to yottabytes, with millions of IOPS per node and throughput into the low hundreds of GBps. On the software side it runs IBM’s Spectrum Scale / General Parallel File System and can give object access and tiering via OpenStack Swift and S3 APIs.

It is targeted at unstructured data and AI/ML workloads and for deployments that can build clusters across wide geographical areas. Storage media supported include spinning disk HDDs and NVMe SSDs. Connectivity is via 100Gbps Ethernet and Infiniband.

IBM Storage Fusion integrates compute, storage and networking into a hyper-converged infrastructure (HCI) system that comes with the Red Hat OpenShift Kubernetes application platform. Fusion also supports S3-based object storage on-prem or in the cloud and customers manage Spectrum Fusion through a single management portal.

IBM is also a big player in tape, with tape, tape management and virtual tape library products up to the latest LTO-9 standard in its TS series range.

Many IBM storage products are also available as software, including Storage Fusion, Storage Scale, Storage Ceph (block, file and object storage aimed at AI workloads), Storage Insight AIOps, Storage Protect cloud-to-cloud data protection, and Storage Defender which can combine with Storage Protect, FlashSystem’s Safeguarded Copy, Storage Fusion and Cohesity’s DataProtect. IBM Storage Sentinel is the company’s ransomware discovery and recovery product for its Oracle, SAP HANA and Epic healthcare products.

IBM Spectrum Virtualize – formerly SAN Volume Controller or SVC – and a venerable part of IBM’s range, seems to have retained the Spectrum branding. It is a block storage virtualisation system that can allow storage hardware from multiple vendors to talk to each other as well as to storage in the cloud. It also offers advanced storage services such as snapshots, replication and storage tiering.

What markets and workloads does IBM target?

IBM is huge, with a very broad range of storage products. It has the capability to deliver storage for any workload. As with other IT and storage vendors, key focuses currently are on flash-driven hardware, AI-based use cases, increased use of the cloud, and the rise of cloud-native applications and containerisation.

How does the cloud fit IBM strategy?

IBM has bet heavily on the cloud as the future for compute and storage, in hybrid- and multicloud modes. Key to this strategy is use of Red Hat Ceph and OpenShift as a platform that leans heavily on use of containers and cloud native applications in cloud environments.

The company is perhaps unique among array vendors in having its own hyperscaler-style public cloud with services for block, file and object access modes. In addition, its hardware and storage software products also allow for hybrid cloud operation. 

What is the IBM container strategy?

In keeping with IBM’s strategic orientation to the cloud following the acquisition and assimilation of Red Hat, containers are a key element. Containers can be managed via Red Hat OpenShift or IBM Cloud Kubernetes service.

Red Hat OpenShift Container Platform delivers orchestration and management for Kubernetes clusters across the enterprise. IBM Cloud Kubernetes service can create and manage Kubernetes clusters on IBM cloud.

What consumption models of purchasing does IBM offer?

IBM’s storage as a service works across on-prem datacentre and cloud and is based on IBM FlashSystem and DS8K hardware. It comes with a base level set at current needs plus 50% extra pre-installed. Base and expansion capacity cost the same.

Storage Utility is a pay-per-use model that delivers 200% more than base capacity from day one. That means datacentre upheaval is avoided by over-provisioning and then using IBM Storage Insights to monitor further capacity needs.

Customers pay only for what they use and if their data needs shrink during any month, the bill will reflect usage, albeit with a minimum “base”.



Source link