Hi AMMRL!
I’ve recently started managing the Chemistry facility at University of Chicago, filling the vacancy of Antoni Jurkiewicz, who retired. I’m happy to be back participating in this community.
I’d like your help improving our computer network architecture and devising a protocol for safely and easily getting user data from spectrometer computers to user computers.
I recently had a discussion with our very helpful and knowledgeable university IT staff, and realized we could make major improvements in usability, efficiency, and security using today’s technology, which I’m frankly not that familiar with. We have an opportunity here to ask “If we were to set up this system from scratch today, how would we do it?” Of course, I’m interested in learning from you all what you’ve got working, what you’ve tried, what you’re happy/unhappy with, etc. We know that migrating a facility system laden with history and habit has challenges different than starting from a clean slate, so your insights on migration strategy are welcome.
I’d like to know how your computers are connected together and what your procedure is for getting data to the users. I’m especially interested in hearing about Chemistry facilities (whose users don’t usually have the scripting/command line skills more common in biomolecular NMR) and industrial facilities (where security concerns are important).
I’d like any information, anecdotes, stories, etc. you have, but to get you started, perhaps address:
- Are your systems running Windows, Linux/CentOS/Ubuntu, or a mixture of both?
- Are your NMR systems connected in a Local Area Network (LAN), so they can talk to each other in lab, but not to the outside world?
- Are the computers on the spectrometers directly accessible from the outside world? Can you get to them via an intermediate computer? Can you operate them remotely?
- Can you access the outside world (internet) from the spectrometer?
- Do you use any high-speed networking technology? 1GB/s? 10GB/s? How do you use it?
- Do you use institutional servers outside the lab for data storage and access?
- Do you use cloud storage such as Box or Google Drive?
How are the user accounts set up?
- Does each user gets their own computer account?
- Do all users share one “walkup” account on the spectrometer, but log in to an individual account within the software (like in IconNMR or (bless it) VnmrJ)?
- Do all users in a research group share one computer account?
- How are privacy/access permissions set up? Can users see each others’ data?
- Do you use any Bruker Topspin tools for walkup access control and/or accounting?
- Something else? Combination?
I’ll stop there before getting too prescriptive.
I’ll happily compile and summarize your responses and report back. After anonymization, I’ll be sharing this info with our networking IT person, and I’ll let you know later what solution we’re thinking of.
Thanks. - Josh
*Our systems, for example:
- We have two labs in separate buildings, and each is on its own LAN. All systems are on CentOS Linux, but there’s currently version heterogeneity among computers.
- We have five Bruker instruments (one with automation driven by IconNMR), two offline processing computers, and several computers for networking and backup.
- All users have individual Linux accounts, so each computer has ~400+ user accounts. On all manually-operated instruments, users start by logging in to their Linux account. On the automated instrument, Topspin/IconNMR is run in the nmrsu account, and users log in to Icon; users accounts are created by making Linux accounts first, then activating them in Icon.
- *Billing is handled on manually-operated instruments simply by reading the Linux login/logout records. On the automated instrument, we use Icon’s Accounting tool (which needs a little help, in my opinion).
- Users cannot log in to spectrometer computers directly from the outside world (home, lab, etc.). But in a terminal window, they can access the data transfer computer, then access the command line environment of each spectrometer.
- Data collected on each spectrometer must be transferred to a separate computer that can be seen from the outside world. There is one such “data transfer” computer for each LAN.
- In our existing protocol, users are taught how to open terminal windows and use Linux commands to copy data from spectrometers to the central data transfer computer. This is a major pain point, as most of our new users today have never even seen a command line interface before, and the jaws of some actually drop in disbelief as they come to terms with the protocol. Even with experienced students and postdocs, this is the only part of their lives that involves a command line. With no other reason to learn anything about Linux, most simply memorize a series of commands someone in lab tell them about, and they just type them to get their data and move on with their lives without knowing what the commands do. Lately I’ve tried educating them on how to create links on their spectrometer desktops so they can drag files graphically from one computer to another, but interest in that has been limited. Surely there’s a better way.
- Once data is on the data transfer computer, users can copy it to their laptops using a free GUI-based application like Filezilla.
- One confusing quirk of our system: The apparent name of our data transfer computer is different whether you’re inside the LAN or outside. Within the lab, the data transfer computer is named something like “nmrA”. From the outside world, users cannot access “nmrA” directly - they must seek access to “nmrB”, which is a separate computer that automatically moves them them to nmrA. There’s a separate nmrA/nmrB pair in each of our buildings, and each of the four computers is named uniquely. I’m sure this system has kept us very secure, but explaining it to new students is problematic.
P.S. In case you look up this facility, please note I’m setting up a new website from scratch, but it’s not live just yet. Most of the information on the existing site is correct, but improvements are on the way.
Received on Thu Nov 21 2019 - 06:18:18 MST