IBM’s entrance into the converged infrastructure space is not “me too” according to the company. While IBM touts many of the same benefits we’ve heard before from time-to-market leaders like VCE (Vblock) and Oracle (Exadata), IBM claims it has differentiation that, while nuanced, creates substantial value for customers beyond the competition. This differentiation includes maximum choice, packaging from a clean slate perspective and IBM’s deep expertise, integrated in a system. This was the message put forth this morning by IBM’s executives at a private analyst meeting at the Andaz Hotel in NYC.
What follows are some notes and highlights from the sessions.
Nancy Pearson – VP Expert Integrated Systems Marketing
Expert Integrated Systems – new category
Global launch – 20,000 people in 6 different locations – virtual broadcast to 8,000+ people
Most effort goes to maintaining existing infrastructure – ~30% goes to innovation.
IBM commissioned study to quantify the anatomy of project deployments – Below is a really crappy picture of the results you can squint through and see the data:
Typical IT project time and budget:
Specify/Design etc – see pix
Cluster & HA
Causes of project delay:
Troubleshooting, tuning & production
Integration, configuration, testing and infr.
Installation, cabling & nw access
Integration, config & testing of apps
Integration, config and testing of mw
Config, build and deploy apps
My PoV: IBM and other suppliers talk about how most of the investment goes to keeping the lights on versus innovation project. Much of this problem comes from the processes between development and operations and this is where the industry – IBM and others need to really push the envelope. The answer is Devops and converged infrastructure supports the trend but it is not the be all and end all answer for users.
IBM’s answer – Expert Integrated System
*Integration by design
“Not a vendor Lock-in approach – it’s inclusive”
My PoV: I love when vendors say there’s no lock-in…converged infrastructure is the mother of all lock-ins. The more integration, the more lock-in (think iPhone). Users should understand this but as well, understand that there’s clear value in integration, for which you will pay a premium, which is fine when justified—often integration, while increasing capex will pay off huge in opex savings. The key issue for users to understand when comparing systems is there is a range of openness from single stack to more heterogeneous.
Today announcing 2 new family members –
1) PureFlex – infrastructure system – expert at sensing and anticipating resource needs to optimize inf.
2) PureApplication – expert at optimally deploying and running apps
Ambuj Goyal – GM dev and mfr IBM
Those familiar w/Exadata, VCE, UCS, etc might say IBM too little too late.
Keep three things in mind:
1) Choice (and flexibility) – it’s hard to do flexibility and integration at the same time.
2) Clean Slate – if you were to bundle you’d spend $150M on a project and take 1.5 yrs to implement – this was 3.5 yrs in the making – security built in from the ground up, mix and match architectures, secure boot, secure firmware install, secure resource sharing between Intel power and nw – investment of $2B?
3) Expert Integrated Systems – ea vendor is limited by its expertise – IBM $6B in R&D ea yr. 10,000’s of deployments ea. Year – have captured significant expertise for tedious things like firmware upgrades and lifecycle mgt as well as optimization of licenses, storage, networking usage, etc.
Pre-configured, pre-integrated, server, storage, networking – automated provisioning, so what’s new?
Existing data centers do not need to be forced to migrate. May be running Windows, SAP, Database on Unix, analytics. We can migrate everything to vmware or PowerVM and consolidate – that’s new – hard to give choice and flexibility – cloud ready w/SmartCooud, Hypervisors treated as workloads, any combo of VMware, KVM, HyperV, and PowerVM…IBM claims it is the best, any combo of aix, ibm i, windows, linux from redhat and suse.
$1/2 of acquisitions such as BNT for “scale in” networking – needed a bus or channel like capability that is efficient. Needed leverage from acquisitions on the sw side.
Challenge – how to fit into existing data center. Ex – I’ve spent money on existing storage – ibm or non-ibm. Use SVC through FlexSystems Manager – EasyTier technology optimizes placement of data.
Innovation – multi-threading, 10gigE, tiering, interconnect, flash, etc…”and we bring to the client one piece at a time.” Then over time the infr becomes inflexible, tedious and slow to change. How do we create the environment so that these innovations can be consumed and adopted more quickly and easily.
Consumability of innovation becomes “a bridge too far.”
Car anology – open the hood before you byu and you’ll see the difference b/tween an engineered system from a clean slate versus a bolt on.
PureFlex lifecycle management – does the tedious things well – e.g. setup, element discovery, firmware upgrades, Chassis maps, integrated open fabric manage, reliable logging / events
Built-in expertise – workload placement service, features on demand, system pools mgt, vmcontrol, storage & NW control.
IBM treats everything as a workload and flexsystems manager takes care of managing the flow.
Built on SVC and V7000 –differentiator for IBM is virtualization of installed heterogenous assets – which will facilitate migration.
My PoV: This is a differentiator from Oracle and EMC for sure. Others such as HDS have this capability but no doubt with respect to the converged infrastructure leaders, IBM is really the only one touting the ability to use virtually any storage. In reality I see this as a migration capability that provides a path for customers to get from point A to point B without rip and replace – which is good.
Networking – Ethernet, FC, IB – can assign vlans and move them appropriately. Outside the system is standard cisco, brocade, etc but inside is “scale in” ibm IP – acquisition of BNT enabled.
Key message is consolidation w/o migration. Not forced. If someone wants to drop the workload into a VMware pool and not use IBM’s systems manager – IBM’s response is “no problem.” The networking objective is to give internal networking capability – like a high bandwidth bus. Outside the system it’s standard.
My PoV: The messaging is good from IBM, however the value proposition it is very much similar to those we hear from VCE, HP, NetApp and Dell. Oracle has major differentiation with built in database and apps and I expect to hear more about that from IBM when it discusses its PureApplication offering. The bottom line is IBM is entering the game in earnest and focusing on what you’d expect – IBM’s expertise at integration and workload knowledge. In my view, IBM is doing a very good job of balancing the need for flexibility and openness with integration and speed to deployment, leveraging its vast portfolio. This announcement will be very competitive. IBM is protecting its large systems and software installed base and this announcement is a productive result of IBM’s re-organization that put the servers, storage and software groups under one roof.