Solid-state storage not there yet, says Cern
Advances in computer storage are being ignored in favour of 'tried and trusted' solutions, according to a network administrator at the Cern particle physics laboratory in Geneva.
David Foster, the head of communications and networks at Cern, explained that the new technologies have not filtered down to his facility's modern networks.
"Tape is still the most effective way of storing large volumes of data," he said. "But it is a clumsy old technology that has its own problems."
One of these problems is that data has to be moved from disc to network and then back onto disc.
"These systems are slow and are not progressing particularly quickly, and solid state memory and solid state storage are important things to look at," said Foster.
Installing fast networks is not enough when it comes to handling data, and it is the end-to-end experience that is important.
"We are always looking at how to transport data effectively over long distances," said Foster. "TCP/IP in its naked form is not the best transport for doing this. You need to modify it."
Cern is collaborating with other research institutes to try and address these problems. "CalTech is a particular example that looks at a lot of these networking improvements," said Foster.
Advances in computer storage are being ignored in favour of 'tried and trusted' solutions, according to a network administrator at the Cern particle physics laboratory in Geneva.
David Foster, the head of communications and networks at Cern, explained that the new technologies have not filtered down to his facility's modern networks.
"Tape is still the most effective way of storing large volumes of data," he said. "But it is a clumsy old technology that has its own problems."
One of these problems is that data has to be moved from disc to network and then back onto disc.
"These systems are slow and are not progressing particularly quickly, and solid state memory and solid state storage are important things to look at," said Foster.
Installing fast networks is not enough when it comes to handling data, and it is the end-to-end experience that is important.
"We are always looking at how to transport data effectively over long distances," said Foster. "TCP/IP in its naked form is not the best transport for doing this. You need to modify it."
Cern is collaborating with other research institutes to try and address these problems. "CalTech is a particular example that looks at a lot of these networking improvements," said Foster.
0 comments:
Post a Comment Subscribe to Post Comments (Atom)