Post Denmark, the national postal service in Denmark, is seeking financial compensation from IBM after a bug in the DS8000 storage array paralyzed its system for almost a week.
IBM failed to discover the bug, which caused the DS8000 to spontaneously shut down every 45 days or so, before shipping the product in March.
"We have filed a claim to IBM Denmark because they haven't delivered a product of the expected quality," said Michael Orno, director of IT operations at Post Denmark.
According to an IBM spokesperson, IBM has no record of a claim filed by Post Denmark for financial compensation to date.
"IBM contacted us on a Friday and told us the system would go down on Tuesday unless they could update the microcode," Orno said. "This was very short notice. It was expensive as we had to get people to work over the weekend."
IBM patched the problem but in mid-June, the DS8000 crashed again and it took the company three days to get it up and running, according to Orno. IBM eventually pulled the plug on Post Denmark's DS8000s and replaced them with older Sharks, free of charge. The postal service has 120 terabytes (TB) on six Shark arrays.
"We don't intend to power up the DS8000s until IBM provides us with two customers that we can talk to who are running the system in production," Orno said. "It has shaken our confidence in IBM's storage systems … it's unlikely we will buy more from them in the foreseeable future." Post Denmark quickly found another supplier willing to step into IBM's shoes. It has bought two DMX 2000 arrays from EMC Corp.
Orno added: "The most distressing part for us came after the second code upgrade … I asked IBM to be frank and to tell us 'have you had problems in other places with this system?' The general manager in Denmark looked me in the eyes and said 'no.' If IBM had been straightforward, we would have tried to solve this issue together."
Users across the world cite similar problemsBaldor Electric Co. based in Fort Smith, Ark., had one DS8000 in production when IBM called and told the company the system would have to be taken down for several hours to install a microcode update.
"We told them, 'you didn't agree to that upfront,' " said Mark Shackleford, director of information systems at Baldor. "We insisted [the upgrade] had to be concurrent." Baldor runs a 14 TB global implementation of SAP on the DS8000.
On July 17 IBM installed a concurrent update to Baldor's system. "It was about 20 minutes to reset the counter," Shackleford said.
For iT Austria, a service provider in Austria that includes the country's two largest banks as its customers, its DS8000s are not in production yet so the outages have not had a business impact.
"Production still runs on our HDS 9980 and ESS 800T … The only impact the DS8000 troubles really had until now is that testing takes longer and consumes far more resources (staff, time, etc.) than we planned, and the visions we had will still take quite some time to come true," said Karin Pöschel, head of department, central storage management, iT Austria.
IBM confirmed the problem last week and issued the following statement over e-mail: "We have addressed all concerns in regards to the DS8000 brought to us by customers, which is having extremely good market acceptance, and we are unaware of any problems with systems shipping today."