Sears eschews IBM/Oracle for open source and self build
Summary: As Sears refactors its IT landscape, legacy vendors are finding themselves out in the cold in favor of open source alternatives.
In a comprehensive research note from Cowen & Co, analysts Peter Goldmacher and Joe del Callar note that IBM and Oracle are being edged out of Sears in favor of open sources alternatives:
The primary issue facing Sears was the reality that the current line up of technology it had been using from its existing vendors wasn't able to keep up with the demands of the business. The benefit of its transition to a custom environment [that includes significant open source adoption] is a more nimble IT infrastructure that can better support the business at a materially lower cost.
In making this radical shift, Oracle and IBM are the losers:
IBM mainframes and Oracle Exadata were two technologies described as too expensive and not flexible enough to adapt to the changes and seasonality of the business. Oracle RAC is being used only where absolutely necessary, and MySQL is seen as a compelling and viable alternative in most situations. The company is also moving away from Microsoft (Mail), Netezza (DW) and Greenplum (DW) in favor of open source solutions wherever possible.
What of the future?
Sears is investing aggressively in Hadoop, Linux and Private Clouds...these technologies are perceived as more flexible and cost effective. Sears has also been investing in IT personnel with advanced skills. Even with the increased investments in staff, the cost savings associated with getting off legacy technologies are perceived to be well worth it. Standardizing on as few technologies as possible is enabling IT to leverage staff across multiple requirements. ETL workloads on Hadoop have helped Sears make big strides towards a "Single Source of Truth" that was previously impossible. Sears is investing in high bandwidth connections to its stores and mobile, which should increase demand for Big Data mgmt and analytics.
Also see: Sears eyes big data for dynamic pricing, cost savings
Cowen's provided some interesting nuggets in the detailed analysis including:
- Turning IT services into a commercial entity through the MetaScale subsidiary. Sears hopes this will help other large IT orgs adopt open source by leveraging Sears' internally developed skills. (Vinnie Mirchandani has been talking about this trend for some time.) It is developing talent from the inside as part of this exercise and already has customers in healthcare and high speed trading.
- New devices replacing handheld from 5-10 years ago.
- Mainframe giving way to private cloud which is perceived as more scalable at lower cost.
- Upfront investments for multi-channel delivery and loyalty programs are expected to lower future costs.
- Open source removes the traditional relational database dependency.
- Putting petabytes of data into an Oracle stack was deemed 'extremely expensive' and even then Exadata could not handle Sears workloads.
- SAP is ostenisbly interested in integrating with Sears' open source solutions.
- It will take Sears around 2-3 years to move to a private cloud. This time frame is designed to ensure no breakages along the way.
- The general strategy is to be as devoid of vendors as possible, concentrating on open source wherever possible.
While this kind of landscape is commonplace among newer businesses like Google, Facebook and so on, it represents a radical departure for an established business. I can imagine that many eyes will be watching closely to see how this ambitious project turns out.
Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.
Talkback
This has been my view all along
The economics of such solutions will drive them forward in highly competitive sectors. The fact that Sears apparently is becoming an IT service provider to other entities is also interesting.
A few points were confusing to me
Also, wouldn't EnterpriseDB (PostgreSQL-based) be a much better drop-in for Oracle's proprietary relational DBMS than MySQL?
And, SAP? There are also mature open-source CRM and HRM solutions available (e.g., SugarCRM).
lower in cost, perhaps initially...
Staff will be building non-standard homebrew bridges at accomplish what mainstream software provides.
I think in the short term I think this is great. In the long term, its all guessing if it'll play out and be lower in cost.
That's a good point
I don't see what you are talking about.
Linux professiionals, like myself, don't do one off, highly specialized systems with Linux any more often than the same is done with proprietory OSes, for exactly the reason you point out.
It appears Sears plans on using mainstream open source projects, without modification, because the open source projects already comple with their needs.
Sounds like the beginning of a study in IT failure.
That is a risk in ALL IT projects
The very big difference
The paper tiger always has bigger teeth than the real tiger.
However I think build your own is highly viable but you need modern, flexible, reliable, productive, business oriented tools based on a sound theoretical basis - not Hadoop for God's sake.
Not as bas as it sounds
I last used key value pairs
Hadoop still processes data, something that last happened on IBM mainframes running COBOL in the 70s.
Hadoop is interesting as a piece of industrial archeology, but has little or nothing to do with modern data management.
As a further explanation
Unfortunately some people are so conservative they still haven't caught up with the big breakthrough made by Edgar Codd with the relational model.
The amount spent on Oracle/Agile SUPPORT ONLY at our company was
it's about risk.
the good thing about using Open Source is that you have autonomy: you are not beholden to some external entity who may be interested in building your business, or perhaps in building their own business. to the CYA mentality, having a vendor to blame is a huge win, even though it doesn't provide that much risk mitigation. having the source and the inhouse expertise, that's *real* risk management.
DIY is not the same thing as amateur. sometimes the wheel needs to be reinvented. inhouse, opensource projects are a lot like eating your own dogfood.
It's about taking control of critical IT assets with open source
"The general strategy is to be as devoid of vendors as possible, concentrating on open source wherever possible. "
The idea that companies can use open source technologies + internal staff to get more from their IT is something we address in this Openbravo white paper:
http://pages.openbravo.com/rs/openbravo/images/ERP_Legacy_of_Frustration_WP_EN.pdf
The context of this is ERP for mid-sized companies (not mega-retailers like Sears), so may be more relevant for most readers. In any case, the concepts strike me as quite aligned.
Disclosure: I work for Openbravo.
BTW, I agree "Hadoop can be tamed", and can be a big win for large retail. Still lots of batch-style processing that needs to be done in that domain...