lmh wrote:
The thing that is scaring me most is the move towards data being stored in random virtual locations (cloud etc.). These schemes always seem to work extremely well on paper and when you have super-fast hardware that isn't doing anything else. Meanwhile almost every bit of software in the world seems to want to phone home every couple of seconds for some reason or another, so the world is changing:
5-10 years ago, the world was full of super-fast computer hardware waiting (frenetically, cycling pointlessly away) for me to press a key.
Today the world is full of me (and other sad individuals) waiting for even-super-faster computer hardware that's waiting (frenetically, cycling pointlessly away) for a response from the network.
The instruments we've had that have proved most reliable have always been on isolated PCs running exactly what was installed by the manufacturer, no updates, and no proper internet access. Yes to a limited LAN to convey data to some place from which it can be accessed, No to anything cleverer than that. It's getting really hard to work that way, but I've regretted every single instance where an instrument has become dependent on more than its own control PC.
Thinking of phone home software, hasn't Agilent put that into their new Chemstore versions of Chemstation? I remember our service rep telling me that the newest version would phone home to confirm licenses and receive updates. It would be one step up from Shimadzu requiring the thumb drive key to make their software run. I do understand the manufacturers wanting to prevent you from upgrading every system you have when you purchase one new instrument. But then there are companies like Metrohm that pretty much give the software away free if you buy the instrument.
We tried using a cloud based system for data archival here, but the connection was so slow it took days to upload a years worth of data from one GC. We tried two low cost NAS systems and they crashed after about a year and had to be sent off to retrieve the data. We collect the same number of scans per sample on MS data and the equivalent for GC data, yet the file sizes have grown exponentially over the years causing us to need several terabytes of storage for each years data, and we even reduced our data retention policy to 5 years from 7 years, but still run out of storage.
Two things I have tried to push here have been a separate, or at least semi separate network for the instruments versus that for the desk computers, but I get pushback because of cost since it is cheaper for them to simply have us do all of our work on the computer that comes with an instrument. The second was to assign fixed network addresses to every computer on the network instead of using DHCP to auto assign. The Agilent software will not work well using DHCP, and how hard would it be anyway to assign and IP address to each computer when we have less than 100 in the building. I keep hearing that DHCP autoassign is much easier and less hassle. So we still end up with needing two network cards per computer, one for the instrument and one for the house network/LIMS.
From the program point of view win10 should run all the software you have. Only bugs in the software itself can prevent this.
A lot of the older versions of Agilent software wants to access the hardware directly, which W10 does not like to allow, this is what causes much of the incompatibility. I remember when the SP3 update for XP crashed most systems, and Agilent added a version check the MSDChemstation install that would not let it install if you have SP3 on the computer. After that we were blocking all updates, that is until they installed the new server and the person who set it up made it activate updates on all attached computer, greying out the boxes on those computers that would allow us to turn it off. To keep our instruments from crashing during updates we had to go in the back door and disable the update program in the Services menus which has worked so far.
Getting IT people who run the front office to understand that you can't apply every single update from Microsoft to our instruments is frustrating. And when they do apply the update and crash something, they try to blame the instrument companies for it.
