Am I really a workstation Idiot?

[al_bickers]

I really need a bit of advice before spending a bucket of my boss' money, and hope that someone will be willing to do me a big favor and take a look at the specs below. (Hope I've posted in the right forum!)

I'm trying to get a reasonable workstation configuration for some analyst-data crunchers. Their preferred spec comes out at around $10K, but we can afford max $4K per machine. I had to make some major tradeoffs--we will be buying several of these machines, and they all have to get the same rig (otherwise they complain that so-and-so has a better machine).

They called me an "IDIOT" for the spec I've come up with: does anyone care to confirm their allegation or defend me? The problem is they only have a certain amount of money, and some trade-offs must be made as much as I'd like to give them everything they want.

Use case: mostly data analysis (R/SAS/Stata/Python), ESRI GIS, MySQL/Posgre plus some analytical app development (light front-ends, if any in C++, C#, Java). They analyze small to medium-size datasets (never TB-class, and rarely GB-class). No CAD or 3D rendering, light use of Adobe CS.

To their disappointment, I've come up with the following (we must by from HP or Dell, can't build):

Model: HP Z420
CPU: Xeon E5-1650 (6C@3.2GHz)
Graphics: FirePro V4900 (to support their 2-3 1900x1200 screens)
Memory: 16GB DDR3 (4x4GB for option of going up to 32GB using all 8 available slots)
HDD : 256GB SSD boot + 3 x 7.2KRPM SATA 1 TB in RAID 0

They really wanted: 64GB, 3x600GB 15k SAS arrays, a faster E5-2600 series processor and Quadro 4000 graphics.

*The SAS drives are expensive, and I can't find any performance justification for their use case.

*I can't see any benefit to a E5-2600 series processor in a single CPU system. I realize there are more upgrade possibilities if we go with a system that has two CPU slots (e.g. HP z820 or z620), and there would be a marginal benefit of going to 8 cores (but this might be offset by the lower clock speed in the affordable E5-2600 8c CPUs, and we are on a 3yr upgrade cycle anyway).

*They don't do any heavy graphics, so a high-end graphics card is out of the question unless I'm just not understanding.

So, am I an idiot? Have I made the right trade-offs, or do I really just not get it? Any suggestions?

I'm the money guy, not the computer guy (I have a CS degree, but that isn't useful here). It is the end of the fiscal year and if I deny their request, they can't resubmit for another 2 years (and their computer budget will be lost). So, I'm trying to muddle through.

THANKS in advance!

 

[Smoblikat]

Im not sure why you would get a 356gb SSD for boot, im sure windows and all their apps would fit on a 128. As for CPU, I would go with a dual socket mobo. Even if you only have one socket filled now, imagine how much cheaper it will be down the road to just buy everyone a new CPu rather then spending money on all new systems.

 

[al_bickers]

Thanks Smoblikat 🙂

I agree with your logic. However, to get systems with two CPU slots, I have to go to the HP z620 workstation (from the z420, = +$300).

Only the E5-2600 series processors can operate in dual CPU systems. The E5-2650 (8 cores @ 2.0 GHz w/20MB) benchmarks similarly to the E5-1650 (6 cores @ 3.2 GHz w/12MB), and is the cheapest 2600 series that has more than 4 cores and supports 1600 MHz memory (still $500 more though). But I worry about the low clock speed--giving up 1.2 GHz for 2 cores and a bit of cache seems harsh.

Do you think real-world performance of the 1650 and 2650 will really be similar even given the 1.2 GHz difference?

Plus this increases the cost by $800, although you're right that in the long run it could save quite a bit.

 

[slugg]

An SSD just to improve boot times is kinda stupid. Nobody cares about boot times on a workstation - you almost never reboot them! Nix it; it's worthless. Go for a really fast RAID setup of 10k RPM drives. The 15k RPM drives won't really help much. But if you can get the 15k RPM drives for a good deal, then sure.

Data set size is rarely GB class now, but what about 3 years from now, by their next upgrade cycle? My data sets I use in my lab are about 1 GB and I find myself running out of memory VERY quickly. We have 16 gb machines. Max the RAM out to whatever you can. Then again, my work is nothing like yours (computer vision). In my case, data sets grow exponentially over time, simply because sensors (cameras, etc) increase in resolution linearly over each dimension over time, so that's exponential growth. What about your application? Think about that with your application.

Video card: no need for a CAD class video card. You're right on picking the cheapest thing that'll support the massive screen resolutions over 3 monitors.

CPU: That's really dependent on the software your team is using. If it's highly parallel, then go for as many cores as you can. If it's not, then go for the fastest clock rate that you can. In my work, everything starts out as a single threaded design, so initial work is much nicer on a lesser-core machine with a faster clock rate. Eventually, we process things in parallel once the algorithm has been finalized, so many-core setups are better for production/usage.

Hope this helps.

 

[sgrinavi]

Are they using any cuda enabled apps? could be why they want the quadro cards

 

[Smoblikat]

I would say real world performance would favor the 2650. Remember, the 1650 is only clocke that high to be relevant to todays performance. The 2650 can chieve the performance with a drastically lower clock speed, and imagine how much better two of themw ill be. Plus you could try a little overclocking with software if you or some people get bored 😛

Also, If you can do without an SSD go for it. But if you can get one and have all the apps and temp data loaded onto them im sure it will be leagues better than any RAID array simply due to loading times. Less time waiting = more time working, which leads to better productivity.

For memory, see if there is an option to go with 2x8gb so you ca upgrade in the future without having to replace all 4 DIMMS.

 

[unokitty]

So, am I an idiot?

Whether or not you are an idiot, I can't say... But I will say that you seem to be putting yourself into a politically untenable position.

Realize that all organizational resource allocation decisions, such as you are presenting here, have a significant political dimension.

If I understand the dynamics, you have an organizational group that has convinced themselves that they need a $10,000 machine.

And you seem to think that you can present them with a rational that will make them happy with a $4,000 machine.

In this situation, what is the upside for you? Do you get a bonus if they buy the cheaper machine? Do you get a raise? What do you get? Anything? And have you considered the down side?

The down side is that each one of those technical people will hate you. Each one of them will make it their job to find something that those machines won't do, just to prove that you were wrong. And if there are enough of them, and if they have enough time, they will find something. That is the nature of the situation...

Spending your political capital in a situation where you have nothing to win but you can lose by alienating a group of co-workers or being involved with a failed purchase doesn't seem prudent to me.

May I suggest that if you want to make a proposal, give that proposal to someone above you. And suggest that the Manager/Director/Whatever of that technical group earn their money by getting a consensus from his group.

This isn't rocket science.

Have your boss, or his boss, give the group Manager/Director/Whatever the HP, or Dell, or Whatever, catalog and the amount of money budgeted. Tell him he has X time to make a decision and that he can buy whatever he needs as long as it doesn't exceed the budget. If you want, you can have him include your proposal with the catalog.

If he doesn't want to buy a standard machine. Tell him that that is okay. But they have to include X years of onsite service with whatever brand X machine that they want ... And the money for that service comes out of their budget.

Getting in the middle of a political situation like this is unwise. Unless of course, you don't like your job and you already have your resume updated.

You don't have to believe me. But I have been directly responsible for the acquisition of over a million dollars worth of computer and communications equipment. And involved with the purchase of well over double that...

What I have learned is that the process that you employ in making the decision is more important than what you buy... And that the people that will be using that equipment need to have a stake in the successful outcome of that process.

And that an unstructured process is full of land mines for anyone that is in the general area. If that new workstation doesn't work out, everyone involved with the process is going to get hit with the failure stick. Why would you want to take that chance?

Hope that everything works out for you.
Uno

 

[Lifted]

Have your boss, or his boss, give the group Manager/Director/Whatever the HP, or Dell, or Whatever, catalog and the amount of money budgeted. Tell him he has X time to make a decision and that he can buy whatever he needs as long as it doesn't exceed the budget.

If that new workstation doesn't work out, everyone involved with the process is going to get hit with the failure stick. Why would you want to take that chance?
Uno

And what the manager will do is send the requirements to their rep at HP and ask for a proposal that will meet the requirements. If the requirements are merely a list of the latest hardware components with no reason at to why they're needed, then this path obviously won't yield results as the HP rep will simply return a quote for $10k workstations.

If that's the case, ask the analysts for benchmarks, and any other supporting documentation, backing up their stated requirements for these $10k workstations.

There are more steps that can be taken here, but as you have probably surmised, the goal with any project that has a 99% chance of failure is to redirect the responsibility for this failure on anyone but yourself. With a little creativity you can position yourself to take credit for the project if it actually succeeds.

😎

 

[greenhawk]

I had to make some major tradeoffs

They called me an "IDIOT" for the spec I've come up with: does anyone care to confirm their allegation or defend me?

Use case: mostly data analysis (R/SAS/Stata/Python), ESRI GIS, MySQL/Posgre plus some analytical app development (light front-ends, if any in C++, C#, Java). They analyze small to medium-size datasets (never TB-class, and rarely GB-class). No CAD or 3D rendering, light use of Adobe CS.


Model: HP Z420
CPU: Xeon E5-1650 (6C@3.2GHz)
Graphics: FirePro V4900 (to support their 2-3 1900x1200 screens)
Memory: 16GB DDR3 (4x4GB for option of going up to 32GB using all 8 available slots)
HDD : 256GB SSD boot + 3 x 7.2KRPM SATA 1 TB in RAID 0

They really wanted: 64GB, 3x600GB 15k SAS arrays, a faster E5-2600 series processor and Quadro 4000 graphics.

*The SAS drives are expensive, and I can't find any performance justification for their use case.

*They don't do any heavy graphics, so a high-end graphics card is out of the question unless I'm just not understanding.

on the first point, do enough drastic cuts and you might as well not bother. At my work the last round of updates had 1/2 the team handing there laptops back as useless (so a waste of money) as they could not do the critial tasks (run the addon cards, worth more than the laptops).

On the second, without knowing what they currently have and why they want specific hardware in the new setups, I can not say either way. Each thing has it's advantage and disadvantage. It find it frustrating with the local IT when it comes to hardware changes but at least they will allow my choices to be chosen if I can give a reason for wanting that choice (have a laptop for working away from my desk, otherwise head office is cheap desktops).

As to the hardware, performance for SAS drives do not show in most bentch marks as it is seek times the drives are designed for, which helps a lot when using heavy access loads and not just straight read/write speeds. A SSD is better for this job but you sure as heck need a large SSD if they are working on 1TB data sets. Having a small "boot" SSD becomes pointless if the data is not on it.

With the video card, as mentioned, if they need the features that present on the quadro video cards then getting one does make a noticable difference (but not seeing anything in the list that needs a $1000 video card when a basic quadro can still be better than nothing. (assuming they can not re-use the older card).

Personally, to get close to their request I would be looking away from the Xeon range to the business / home gamer range (if you can still get the needed support contracts). Go for a fast 4 core, 32GB ram (max it out), cheapest Quadro you can get (even if a separate extra the users install) and 2x 512GB SSD's. It will probably be better overall for most tasks, but all that depends on what the current systems they are using are. Added benifit is that if a few need more space and some do not need the full 1TB of ssd, then they can give over their spare SSD to the ones that need it (but this should all be done by the users, not your call to worry about).

Otherwise, best option is to give the team leader(?) the hassel. Tell them they have until the end of the week to decide due to the end of the financial year (?, it is in my country anyway), this is the budget set by management, otherwise they go without. Not your issue as you gave a option and the users rejected it (document this and tell your boss so that if the users complain, all is good in your case as you did what you could).