Ibiblio and Linux software storage ?
Ibiblio runs at University of North Carolina which uses RedHat software.
Linux software storage facility had a total of 171 gb of total accumulated data. I don't know if that is all the 18/19,000+ programs.
They have several file servers and go thru Internet clouds then load balancing failsafe arrangement.
In many cases download is done thru mirrors around the world. These mirrors are often changed.
But since Apache is done thru tomcat; the problem is the synchronization with MySQL on data submitted. Sometimes, the storage of data superseded fragmentation limits in the HDD FAT table.
So, they are now planning network drive using virtual servers. This technology change may have cause unusual difficulties synchronizing with other servers. Just a guess, but so far they are still waiting for a solution.
The real problem is the ISP servers matching load balancing schemes with virtual servers? Secured connections often lock up keyboard and mouse until ISP servers releases the lock up.
The problems of using virtual servers is that PcLos website can not take the load redirected by ibiblio traffic. Novell will have hardware problems, once they deploy virtual servers. SunMicro can always support images with their own mainframe computers everywhere that they designate to be virtual servers.
- Login or register to post comments
- Printer friendly version
- 865 reads
PcLos must call Ibiblio at UNC ?
Since the university people at North Carolina did not fully understand virtual servers owned by others, they overloaded these unknown servers and got disconnected themselves without warning.
U of NC must support any overload of virtual servers by mirrors? Ibiblio might have hacked the host server to make virtual server work better and should pay for the damages done.
Puppy is lucky not rich enough to have a hosted server for ibiblio to use as a virtual server?
Glad Ibiblio went back up without virtual servers ?
Ibiblio was back up a few days ago. Most of the files may have to be reloaded.
Virtual servers mean you only need redirect scripts. We used packet switching from 350,000 computers to one wafer scale computer to screen contents of packets. This security measure maybe needed in many future website setup.
I am glad that UNC people fixed their 171 gb storage for Linux. It is a small server, no where near terabytes of info. So why bother with virtual servers, just get more broad band services to handle the traffic. Multicore cpu is a most to handle more small HDDs for faster distribution of indexed data. I personally like 4/10GB 1.0" HDDs or USB flash(!) drives and many of them in parallel to distribute critical data faster and simultaneously..