what did you learn today? (part 2)

[Whittey]

afidel[/url]":3txnnwkq]Yeah, we get 6 hour call to repair on all our HP gear, doesn't cost that much for the first 3 years of uplift. Be aware that there's a clause that gives them an out during the first 90 days of availability of a new model due to normal parts constraints. With hypervisors and live migration I'm not sure it's really necessary anymore but it's a cheap insurance policy in case some freak event should cause us to lose a significant number of hosts at the same time, when downtime is north of $30k an hour you can buy a LOT of warranty uplifts before coming close to even a partial day event.

And there's some other fine print for 6CTR as well. At least for the XP line, the 6 hour call-to-repair does NOT include software issues, performance issues, or disk replacements.

[edit]And we dropped 6CTR on our XP's this year for P24. That was about a 27% drop in maintenance cost.

 

[kennedye]

Man, you guys must be working with a different Dell than we are; the last couple of times we've called them with a bad DIMM it's taken a couple of weeks to get a replacement shipped, plus we have to go through the whole "oh you need to update the firmware and run these diagnostics and blah blah blah" dong and dance before they'll even discuss shipping stuff out.

 

[sryan2k1]

Do you not have Pro / Gold support? Those guys don't ask a whole lot of questions, and generally phone calls are under 15 minutes including the time to open the ticket. If you open it online it might be a 5-10 minute call.


Edit: And yeah, Pro / Gold support is a different internal support group then the normal warranty stuff, so we are dealing with a "Different" Dell.

 

[theevilsharpie]

kennedye[/url]":1m9opigv]Man, you guys must be working with a different Dell than we are; the last couple of times we've called them with a bad DIMM it's taken a couple of weeks to get a replacement shipped, plus we have to go through the whole "oh you need to update the firmware and run these diagnostics and blah blah blah" dong and dance before they'll even discuss shipping stuff out.

I've been through that routine, but at least Dell eventually ships you the parts.

I still remember a few years back when a customer we brought on board was screwed out of warranty service by HP because the customer purchased a Smart Buy Proliant Server with a standard care pack. Apparently, Smart Buy ProLiants can only be tied to Smart Buy Care Packs, not standard Care Packs.

(For the non-HP folks, Smart Buy is HP terminology for a product with a special sale price, and Care Pack is their terminology for the extended warranty tiers.)

I can somewhat understand HP wanting to reserve spare parts for higher warranty tiers (particularly for parts that rarely fail like motherboards), but HP's Smart Buy Care Pack is complete bullshit. The restriction is non-obvious to anyone but the most hardcore HP server nerd, none of HP's other business units (or any other server vendor, for that matter) have that limitation, and HP never raised any fuss when the Care Pack was registered.

 

[afidel]

theevilsharpie[/url]":2btgjgyp]

kennedye[/url]":2btgjgyp]Man, you guys must be working with a different Dell than we are; the last couple of times we've called them with a bad DIMM it's taken a couple of weeks to get a replacement shipped, plus we have to go through the whole "oh you need to update the firmware and run these diagnostics and blah blah blah" dong and dance before they'll even discuss shipping stuff out.

I've been through that routine, but at least Dell eventually ships you the parts.

I still remember a few years back when a customer we brought on board was screwed out of warranty service by HP because the customer purchased a Smart Buy Proliant Server with a standard care pack. Apparently, Smart Buy ProLiants can only be tied to Smart Buy Care Packs, not standard Care Packs.

(For the non-HP folks, Smart Buy is HP terminology for a product with a special sale price, and Care Pack is their terminology for the extended warranty tiers.)

I can somewhat understand HP wanting to reserve spare parts for higher warranty tiers (particularly for parts that rarely fail like motherboards), but HP's Smart Buy Care Pack is complete bullshit. The restriction is non-obvious to anyone but the most hardcore HP server nerd, none of HP's other business units (or any other server vendor, for that matter) have that limitation, and HP never raised any fuss when the Care Pack was registered.

That's when you talk to the local depot manager and your account manager and tell them you're never buying another piece of HP gear if they don't make it right. Hell, it doesn't even make any sense since the SmartBuy prices are generally lower than the standard carepacks, why would they screw you over for spending more money?

 

[theevilsharpie]

afidel[/url]":rm8k73e9]

theevilsharpie[/url]":rm8k73e9]

kennedye[/url]":rm8k73e9]Man, you guys must be working with a different Dell than we are; the last couple of times we've called them with a bad DIMM it's taken a couple of weeks to get a replacement shipped, plus we have to go through the whole "oh you need to update the firmware and run these diagnostics and blah blah blah" dong and dance before they'll even discuss shipping stuff out.

I've been through that routine, but at least Dell eventually ships you the parts.

I still remember a few years back when a customer we brought on board was screwed out of warranty service by HP because the customer purchased a Smart Buy Proliant Server with a standard care pack. Apparently, Smart Buy ProLiants can only be tied to Smart Buy Care Packs, not standard Care Packs.

(For the non-HP folks, Smart Buy is HP terminology for a product with a special sale price, and Care Pack is their terminology for the extended warranty tiers.)

I can somewhat understand HP wanting to reserve spare parts for higher warranty tiers (particularly for parts that rarely fail like motherboards), but HP's Smart Buy Care Pack is complete bullshit. The restriction is non-obvious to anyone but the most hardcore HP server nerd, none of HP's other business units (or any other server vendor, for that matter) have that limitation, and HP never raised any fuss when the Care Pack was registered.

That's when you talk to the local depot manager and your account manager and tell them you're never buying another piece of HP gear if they don't make it right. Hell, it doesn't even make any sense since the SmartBuy prices are generally lower than the standard carepacks, why would they screw you over for spending more money?

My details of the incident are a bit hazy since it happened several years ago. IIRC, the server group tried to vouch for me, but the group responsible for the care packs refused to cover the repair. After nearly 24 hours of back and forth, the best HP could offer me was an on-site service call for $1,500. Also, this care pack actually was a 6-hour call-to-repair.

(As an aside, does HP/Dell/whoever actually have any real liability if they fail to meet their warranty repair time commitments?).

I had the same thought as you: I figured that HP separated the Care Packs in this way because they were subsidizing their server sale prices with higher-priced Care Packs. Nope, the Smart Buy Care Packs were cheaper. And HP acknowledged that the Care Pack was registered and was still within our coverage period, they just didn't give a fuck.

We definitely started courting Dell, but Dell's business model at the time was targeted at sales to end users, not VARs. We were essentially married to HP, and HP knew it

 

[kperrier]

It can be a pain in the ass, but this is why you always have 2 vendors for everything.

 

[antiwraith]

stash[/url]":1fnedrnj]Carpeted floor tiles are awesome, because you get to use this:

I have one of those!

I also have carpet in the server room.

 

[M. Jones]

BitPoet[/url]":2amko0lp]Carpeted floor tiles exist in data centers.

I'm tempted to use these in the future for aesthetics and a bit of noise abatement. Not only will they offer evidence that someone brought coffee into the raised floor area against policy and spilled it, but they will also soak up said coffee and prevent it from leaking down between the tiles. How can you argue with that logic?!

 

[M. Jones]

I spoke with Tom Limoncelli at an event last night and asked him about the prospects for a third edition of The Practice of System and Network Administration. I was a bit surprised but delighted when he told me that such a thing had just been announced the previous day:

"The Practice of System and Network Administration" (TPOSANA) is now 7 years old (with some chapters virtually unmodified since the first edition, 12 years ago). We are preparing to update the book and create a 3rd edition but we need your help!

We're looking for your input! Yes, you! Our valued readers!

We're re-reading all 1,100 pages to find parts that are obsolete or need updating but we need your help. We've decided to crowd-source this part of the project. You are probably a better judge of what is missing, obsolete, or needs updating. Pick a chapter, a page, or a section and file bugs against any issues you find. They can be specific errors ("Typo on page 123") or general ("Chapter 28 focuses on the old Solaris way; add the newer Linux way.")

Tom also revealed something that hasn't been announced yet: that there will be a Volume 2 focusing on cloud computing.

 

[antiwraith]

M. Jones[/url]":3efdy7ni]I spoke with Tom Limoncelli at an event last night and asked him about the prospects for a third edition of The Practice of System and Network Administration. I was a bit surprised but delighted when he told me that such a thing had just been announced the previous day:

"The Practice of System and Network Administration" (TPOSANA) is now 7 years old (with some chapters virtually unmodified since the first edition, 12 years ago). We are preparing to update the book and create a 3rd edition but we need your help!

We're looking for your input! Yes, you! Our valued readers!

We're re-reading all 1,100 pages to find parts that are obsolete or need updating but we need your help. We've decided to crowd-source this part of the project. You are probably a better judge of what is missing, obsolete, or needs updating. Pick a chapter, a page, or a section and file bugs against any issues you find. They can be specific errors ("Typo on page 123") or general ("Chapter 28 focuses on the old Solaris way; add the newer Linux way.")

Tom also revealed something that hasn't been announced yet: that there will be a Volume 2 focusing on cloud computing.

That is awesome! Now to find a copy of the book. I see it is on Amazon for $38 and $34 Kindle. Any idea where I can find a cheaper copy? I don't want to put down $35+ for a book that will soon be the old outdated copy. :bigdumbgrin:

 

[Danger Mouse]

Danger Mouse[/url]":3jakkpqr]...that inevitably, when there's an issue with the network setup in a building, it falls back to the contractor having screwed up the config a bit.

In this case it was DHCP options for a WAP setup.

And now, thinking about it, I think I know why some of my phones are screwing up, because they're using the same DHCP pool for cisco VOIP phones as well as WAPs.

I'm thinking there may be some conflicting DHCP options at work here.

EDIT: Yup, option 60 can be set for only one device per dhcp scope/pool/network (depending upon what you're using). Same for option 43, although option 43 can have more than one IP address in hex, it can be an issue still.

Fascinating stuff. And now I see that the previous laborious manual WAP configuration done by the network staff left at the contractor, are due to the incorrect setup of the dhcp pools and respective subnets and vlan assignments for switchports.

Okay, now I know why it was done that way previously.

That's right. Licenses

The durn WiSM blades were purchased, but not enough licenses went with it.

Mind you, we probably blew 10x as much the licenses cost, on having a contractor configure the access points prior to installation.

 

[flameboy]

antiwraith[/url]":3aw9acmi]That is awesome! Now to find a copy of the book. I see it is on Amazon for $38 and $34 Kindle. Any idea where I can find a cheaper copy? I don't want to put down $35+ for a book that will soon be the old outdated copy. :bigdumbgrin:

That is about as cheap as you can find it. I wouldn't worry too much, it will take quite a while for the new version to come out. Even the 1st edition is still a great book, and most of the advice is timeless.

 

[Scotttheking]

M. Jones[/url]":2qfbottw]I spoke with Tom Limoncelli at an event last night and asked him about the prospects for a third edition of The Practice of System and Network Administration. I was a bit surprised but delighted when he told me that such a thing had just been announced the previous day:

"The Practice of System and Network Administration" (TPOSANA) is now 7 years old (with some chapters virtually unmodified since the first edition, 12 years ago). We are preparing to update the book and create a 3rd edition but we need your help!

We're looking for your input! Yes, you! Our valued readers!

We're re-reading all 1,100 pages to find parts that are obsolete or need updating but we need your help. We've decided to crowd-source this part of the project. You are probably a better judge of what is missing, obsolete, or needs updating. Pick a chapter, a page, or a section and file bugs against any issues you find. They can be specific errors ("Typo on page 123") or general ("Chapter 28 focuses on the old Solaris way; add the newer Linux way.")

Tom also revealed something that hasn't been announced yet: that there will be a Volume 2 focusing on cloud computing.

NICE!

 

[akro]

Whittey[/url]":sm9b6ccj]

afidel[/url]":sm9b6ccj]Yeah, we get 6 hour call to repair on all our HP gear, doesn't cost that much for the first 3 years of uplift. Be aware that there's a clause that gives them an out during the first 90 days of availability of a new model due to normal parts constraints. With hypervisors and live migration I'm not sure it's really necessary anymore but it's a cheap insurance policy in case some freak event should cause us to lose a significant number of hosts at the same time, when downtime is north of $30k an hour you can buy a LOT of warranty uplifts before coming close to even a partial day event.

And there's some other fine print for 6CTR as well. At least for the XP line, the 6 hour call-to-repair does NOT include software issues, performance issues, or disk replacements.

[edit]And we dropped 6CTR on our XP's this year for P24. That was about a 27% drop in maintenance cost.

Typically customers dont even know they need a disk replaced...ctrack should place the order for the new disk fairly quickly after a SIM is generated. Then they just show upwith a disk...

 

[akro]

theevilsharpie[/url]":21scwtjq]

afidel[/url]":21scwtjq]

theevilsharpie[/url]":21scwtjq]

kennedye[/url]":21scwtjq]Man, you guys must be working with a different Dell than we are; the last couple of times we've called them with a bad DIMM it's taken a couple of weeks to get a replacement shipped, plus we have to go through the whole "oh you need to update the firmware and run these diagnostics and blah blah blah" dong and dance before they'll even discuss shipping stuff out.

I've been through that routine, but at least Dell eventually ships you the parts.

I still remember a few years back when a customer we brought on board was screwed out of warranty service by HP because the customer purchased a Smart Buy Proliant Server with a standard care pack. Apparently, Smart Buy ProLiants can only be tied to Smart Buy Care Packs, not standard Care Packs.

(For the non-HP folks, Smart Buy is HP terminology for a product with a special sale price, and Care Pack is their terminology for the extended warranty tiers.)

I can somewhat understand HP wanting to reserve spare parts for higher warranty tiers (particularly for parts that rarely fail like motherboards), but HP's Smart Buy Care Pack is complete bullshit. The restriction is non-obvious to anyone but the most hardcore HP server nerd, none of HP's other business units (or any other server vendor, for that matter) have that limitation, and HP never raised any fuss when the Care Pack was registered.

That's when you talk to the local depot manager and your account manager and tell them you're never buying another piece of HP gear if they don't make it right. Hell, it doesn't even make any sense since the SmartBuy prices are generally lower than the standard carepacks, why would they screw you over for spending more money?

My details of the incident are a bit hazy since it happened several years ago. IIRC, the server group tried to vouch for me, but the group responsible for the care packs refused to cover the repair. After nearly 24 hours of back and forth, the best HP could offer me was an on-site service call for $1,500. Also, this care pack actually was a 6-hour call-to-repair.

(As an aside, does HP/Dell/whoever actually have any real liability if they fail to meet their warranty repair time commitments?).



I had the same thought as you: I figured that HP separated the Care Packs in this way because they were subsidizing their server sale prices with higher-priced Care Packs. Nope, the Smart Buy Care Packs were cheaper. And HP acknowledged that the Care Pack was registered and was still within our coverage period, they just didn't give a fuck.

We definitely started courting Dell, but Dell's business model at the time was targeted at sales to end users, not VARs. We were essentially married to HP, and HP knew it

That is not correct and the DM for your customer should have corrected the issue. Ihave neverheard of a smartbuy carepack versus a regular carepack.

Also for firmware for HP as long as your on an Suport pack for proliant that is less than a year old the only time you should ne told to upgrade drivers or firmware is because your specific issue is known to be resolved by a fspecific firmware/driver update. If you have adrive failure or some other failure they want say update and call me back.

 

[theevilsharpie]

akro[/url]":27pyv20m]That is not correct and the DM for your customer should have corrected the issue. Ihave neverheard of a smartbuy carepack versus a regular carepack.

I was pretty heavily involved in that incident, so if HP no longer makes a distinction between normal and SmartBuy Care Packs, then they did in the past. Also, if you look at Care Pack Central, HP still has a category specifically for ProLiant SmartBuy servers.

I will give HP one thing: to date, I've never been denied a warranty replacement because of an old firmware.

 

[KD5MDK]

I learned that if I email in a service request about the coffee machine, I get a 9 minute response time. (To the email. The actual service call took longer.)

 

[scorp508]

TIL that you can relive the stress of a virus outbreak. I was cleaning my office and came across a large number CDs unlabeled with data on them. Going through them each had a specific set of Windows patches, virus definitions, and cleaners to deal with Code Red, Nimda, I Love You, etc...

I could remember exactly where I was when using each of them and started to sweat a little remembering how crazy those day(s) were.

 

[Thegn]

I remember how ugly things used to get (I was working at Microsoft when some worm got loose inside and trashed a lot of the machines on campus - might have been Nimda. Nobody got anything done for a week or so, although even then I was in a "patch first, ask questions later" mindset so my machines were spared.) but viral outbreaks these days seem minimal. Is it that systems have gotten more secure or that the authors of the various strains of crapware have gotten more subtle?

 

[UruseiY]

the authors of the various strains of crapware have gotten more subtle?

Far, far, far more subtle. Back then it was largely for lulz - now there's a definite slant towards corporate espionage and crimeware where showing signs of your existence are counterproductive.

 

[finni]

InThane[/url]":17fmwlf1]Is it that systems have gotten more secure or that the authors of the various strains of crapware have gotten more subtle?

Pretty certain it's the latter. Today's viruses aren't trying to kill your machine or trash your data or choke your internet pipe, they're trying to copy your data or steal your passwords or mine bitcoins (or over-rev your centrifuges OMG). If they adversely impact the system in a noticeable way, they'll get resolved. if they stay quiet, they can keep doing their nasty business.

That said, I do have one client that keeps getting outbreaks of the virus that hides all folders and replaces them with identically-named files that use the Folder icon in Explorer.

 

[ramases]

Danger Mouse[/url]":3k6q0rxy]

Danger Mouse[/url]":3k6q0rxy]...that inevitably, when there's an issue with the network setup in a building, it falls back to the contractor having screwed up the config a bit.

In this case it was DHCP options for a WAP setup.

And now, thinking about it, I think I know why some of my phones are screwing up, because they're using the same DHCP pool for cisco VOIP phones as well as WAPs.

I'm thinking there may be some conflicting DHCP options at work here.

EDIT: Yup, option 60 can be set for only one device per dhcp scope/pool/network (depending upon what you're using). Same for option 43, although option 43 can have more than one IP address in hex, it can be an issue still.

Fascinating stuff. And now I see that the previous laborious manual WAP configuration done by the network staff left at the contractor, are due to the incorrect setup of the dhcp pools and respective subnets and vlan assignments for switchports.

Okay, now I know why it was done that way previously.

That's right. Licenses

You can have one DHCP Option 43 per host-statement(or client-(sub)class in case you like masochism and how client-class assignment works in ISC dhcpd), and the host statement will override all scope/pool/network definitions for the matching client. You still need at least one scope matching a network interface though, else dhcpd will fail to start with some message about "couldn't find interface to bind to".

Which is also how we run our DHCP setup for APs:
0) Retrieve list of all APs with their MACs/IPs/Hostnames/AP location from network management database
1) Assign APs to certain controllers based on their location
2) Generate one host statement per AP.

Since the AP config doesn't contain any dynamic scopes, we can also run the setup as dual-active without any concern about lease-file mismatches.

The durn WiSM blades were purchased, but not enough licenses went with it.

Mind you, we probably blew 10x as much the licenses cost, on having a contractor configure the access points prior to installation.

m(

Since you are probably speaking about wism2 blades (wism1 blades have no per-AP licensing) you could also look into implementing the HA failover option available since 7.2 (or 7.3, I'm not sure). This would assign a dedicated standby controller to each active controller, and the HA pair can share licences (ie: you can move all licences assigned to the formerly-standalone-now-standby controller to its active partner).

Warning: You cannot do an interruption-free software upgrade with a single HA pair because both members will be unavailable during firmware upgrade. This means an interruption-free upgrade in a HA configuration requires at least three controllers (one active/standby pair and one standalone controller)

KD5MDK[/url]":3k6q0rxy]I learned that if I email in a service request about the coffee machine, I get a 9 minute response time. (To the email. The actual service call took longer.)

The last time our coffee machine was broken I worked from home till they fixed it.

 

[scorp508]

InThane[/url]":85sp63ki]Is it that systems have gotten more secure or that the authors of the various strains of crapware have gotten more subtle?

I think it's a variety of factors.

1. There is more money in turning a machine into a zombie you control than wrecking it. Subtlety wins here.

2. Windows as a target has gotten overall stronger at preventing attacks that do not require end user actions or browsing to become infected.

3. The population is more computer-savvy today than it was a decade ago, making #2 stronger as they are less likely to do as many dumb things.

4. Antivirus efforts are much better with companies sharing information and clients getting better definitions on a more frequent basis.

 

[dlp]

I dunno about 3 or 4, personally. My users still act pretty dumb, and any paid antivirus doesn't seem to be worth much so enterprises get stuck. Heck, my most effective AV seems to be the Barracuda Spam & Virus filter we use.

 

[scorp508]

dlp[/url]":19lamq73]I dunno about 3 or 4, personally. My users still act pretty dumb, and any paid antivirus doesn't seem to be worth much so enterprises get stuck. Heck, my most effective AV seems to be the Barracuda Spam & Virus filter we use.

I said less dumb things, not no dumb things.

 

[Rick25]

That equipment knows when it's replacement is imminent and will begin to fail accordingly.

 

[Technarch]

dlp[/url]":1laxqqhv]I dunno about 3 or 4, personally. My users still act pretty dumb, and any paid antivirus doesn't seem to be worth much so enterprises get stuck. Heck, my most effective AV seems to be the Barracuda Spam & Virus filter we use.

Signature-based antivirus isn't enough any more. Polymorphic malware is getting too common. Sandboxing is better (eg FireEye or Palo Alto Wildfire), but already we're seeing malware that checks to see if it's in a sandbox before detonating. Behavior-based advanced malware detection is currently the best way to go, in my opinion.

 

[mdporter]

Rick25[/url]":24kezgqy]That equipment knows when it's replacement is imminent and will begin to fail accordingly.

++ have one of those right now. Do I do the tickets for the old one, or keep on the death march for the new one and tell them to wait? So it has been the latter.

 

[XOcet]

mdporter[/url]":2404fpmu]

Rick25[/url]":2404fpmu]That equipment knows when it's replacement is imminent and will begin to fail accordingly.

++ have one of those right now. Do I do the tickets for the old one, or keep on the death march for the new one and tell them to wait? So it has been the latter.

Yep, seen that. When a certain SAN vendor's gear fell out of warranty, the next two weeks saw more drives fail than in the previous three years!

 

[Scotttheking]

TIL that apparently it could take a weekend to do a clone of a set of VMs totaling about "a quarter of a terabyte" over 8gig fiber channel links to an EMC VNX and that I really need to get 16gig fiber so it'll only take a few hours...yeah, gotta love this vendor. Stupid reasons mean they can't be removed, but at least I'm doing the design review, even if that means I'm doing their design for them, since I'm the admin who will own it once delivered.

Also, I can't use the storage vendor plugin that does clones directly on the array rather than cloning from the vcenter console because *bad stuff*. Hello, it's a fucking block file copy either way.

 

[Rick25]

VAAI doesn't require a plugin AFAIK, unlike the multipathing plugins for storage isn't it automatically activated if the underlying storage supports it?

 

[Scotttheking]

Rick25[/url]":1dxvjcb7]VAAI doesn't require a plugin AFAIK, unlike the multipathing plugins for storage isn't it automatically activated if the underlying storage supports it?

Not sure, I'm not a storage person (sadly, because I found a neat sounding job but storage is one of the requirements). My quick searching online said there's a vcenter plugin for EMC, but it wasn't super clear, and I didn't research much.

Either way, looks cool: http://www.emc.com/collateral/hardware/ ... vnx-wp.pdf

 

[jshiplett]

VAAI will be automatically enabled with your VNX LUNs. EMC vCenter plugin not required.

 

[ferzerp]

Yeah, the whole point of VAAI is that it is a standard set of instructions specifically not needing any special drivers for offloading certain storage actions to the array.

I hate that recover point exposed volume images don't do VAAI though (at least in physical access mode, and sym doesn't do virtual mode yet), even when the underlying array does. I understand it, it's obvious, but annoying at the extra time it takes.

 

[jshiplett]

Scotttheking[/url]":3m8tm824]TIL that apparently it could take a weekend to do a clone of a set of VMs totaling about "a quarter of a terabyte" over 8gig fiber channel links to an EMC VNX and that I really need to get 16gig fiber so it'll only take a few hours...yeah, gotta love this vendor. Stupid reasons mean they can't be removed, but at least I'm doing the design review, even if that means I'm doing their design for them, since I'm the admin who will own it once delivered.

I know of a good EMC VAR in the DC area if you're looking...

 

[Scotttheking]

euri[/url]":ueqzr9xw]

Scotttheking[/url]":ueqzr9xw]TIL that apparently it could take a weekend to do a clone of a set of VMs totaling about "a quarter of a terabyte" over 8gig fiber channel links to an EMC VNX and that I really need to get 16gig fiber so it'll only take a few hours...yeah, gotta love this vendor. Stupid reasons mean they can't be removed, but at least I'm doing the design review, even if that means I'm doing their design for them, since I'm the admin who will own it once delivered.

I know of a good EMC VAR in the DC area if you're looking...

I wish. Already own said VNX. This project is being put together by system integrators, and they don't seem to know what they are doing, so I'm "making suggestions" against their design. Dingbats actually suggested "let's get another SAN because we might not get enough performance from the existing VNX" where their array was a Dell MD3600 series. Yeah, NO.

 

[BitPoet]

Scotttheking[/url]":2k9s1d4t]

euri[/url]":2k9s1d4t]

Scotttheking[/url]":2k9s1d4t]TIL that apparently it could take a weekend to do a clone of a set of VMs totaling about "a quarter of a terabyte" over 8gig fiber channel links to an EMC VNX and that I really need to get 16gig fiber so it'll only take a few hours...yeah, gotta love this vendor. Stupid reasons mean they can't be removed, but at least I'm doing the design review, even if that means I'm doing their design for them, since I'm the admin who will own it once delivered.

I know of a good EMC VAR in the DC area if you're looking...

I wish. Already own said VNX. This project is being put together by system integrators, and they don't seem to know what they are doing, so I'm "making suggestions" against their design. Dingbats actually suggested "let's get another SAN because we might not get enough performance from the existing VNX" where their array was a Dell MD3600 series. Yeah, NO.

Yep, I'd wait on the MD series until the Soyuz controllers are GA. They're faster than the VNX on writes, at least, not reads, not sure how they scale with drives, though my numbers are 60 drives on the Soyuz vs. 240 on the VNX.

also: standard disclaimer about benchmarking goes here.

 

[ncrand]

TIL That the IS department thinks it's a great idea to make a major infrastructure change on a Friday, right the End of Business which causes an entire centre to go offline. Some much for Read-Only Friday.

Hellooo overtime.

 

[jshiplett]

First rule of system administration - NO CHANGES ON FRIDAY