|
Tab8715 posted:Has anyone without a degree hit any stumbling blocks? Or actually have been denied a position because they didn't have a degree? The one large company I worked for pretty much had a cap for advancement for me without a degree. Granted that cap would have been director and the pay would have been pretty decent but I wasn't quite ready to settle. Even today I wish I had a 4 year degree.
|
#
¿
Aug 15, 2014 08:48
|
|
|
|
| # ¿ Apr 29, 2024 12:27 |
|
|
Physical installation of equipment and initial configuration. That'll go to partners/manufacturers and occasionally be done in-house but it's rarely off shored just because you typically need to at least be present for initialization. Some companies try to outsource anything and everything but ultimately end up bringing large portions of it back in.
|
|
#
¿
Oct 7, 2014 22:12
|
|
|
SSH IT ZOMBIE posted:Ahh....jesus christ. If they make the blades large enough, and at a good price, we'll probably have to honestly consider this at some point. Which means I'll need to learn UCS after all. I'd think of that as good news. UCS is pretty awesome.. Policy based server management and the tool you get out of the box to manage it isn't half bad either.
|
|
#
¿
Nov 12, 2014 06:42
|
|
|
SSH IT ZOMBIE posted:Like, how the hell does it work with HBAs and storage? Storage gets hooked up to the fiber interconnect, OR, you have storage in the chassis, I think. If it's external, how do you define the WWID and such of the "virtual" hbas? NPIV? If you're askin' I'll answer! The fabric interconnects can connect to your existing FC network via either FCoE or native FC. Modern fabric interconnect ports can be configured as either a native fibre channel port or native ethernet port and all you need to do after that is plug in the right optics. You could plug storage direct into the chassis but you don't have to and I generally don't recommend you do so. Blade WWNNs are defined via pools which can be baked into "service profile templates." When you instantiate a blade it pulls a WWNN out of the pool as needed (and it will follow the service profile when you move it to another blade.) Blade WWPNs are also defined via pools and they can be baked into vHBA templates (which in turn can be baked into service profile templates.) By default the fabric interconnect runs in NPV mode (in brocade parlance this is "access gateway" mode) so your upstream FC switch needs to support NPIV. Everything in UCS can be tied back to a pool and/or a policy. Want to run a specific version of firmware or BIOS for your servers? Create a policy that encapsulates all of these into it. Want to give the QA team direct access to blades to power them on/off and get a KVM console? Easy enough through UCSM RBAC. Here's a link to the python UCSM SDK: https://communities.cisco.com/docs/DOC-37174 We're currently using this to push MAC addresses and UUIDs into provisioning tools. Here's a link to some powershell stuff for UCSM: https://communities.cisco.com/docs/DOC-37154 We're currently using this to actually build a new UCS domain from the ground up by feeding it an excel spreadsheet. I love me some UCS and I've been working with it since UCS manager was in a beta for 1.0 so feel free to ask more questions.
|
|
#
¿
Nov 12, 2014 07:21
|
|
|
You typically have 2 IOMs in the back of each chassis, 1 IOM is connected to 1 Fabric interconnect. So you effectively have A fabric and B fabric with respect to blade connectivity so you should be good there. What's nice is a pair of fabric interconnects could have multiple chassis plugged into it. One pair of fabric interconnects (depending on model) could manage say 160 blades for example and you can manage up to 4 pairs of fabric interconnects at no cost using UCS Central. That's basically one point of management and policy enforcement for 500ish blades in your environment. Regarding redundancy: I'm assuming you're looking at getting VIC 1240 or VIC 1280 cards in your blades (or as MLOM.) These are basically cards that let you carve up whatever network adapters and fibre channel adapters you may need. For example my ESXi hosts have 6 network adapters and a pair of FC adapters. Yeah they share the same 2 physical uplinks but it lets me use a standard vSwitch for management, a distributed vSwitch for my guests and iSCSI and a pair of uplinks are left over for my Nexus 1000v testing. When you define your vHBAs you can pin them to either fabric for redundancy sake and you can continue to use whatever multi-path software you like (mpathd, powerpath, whatever.) You'll create a vHBA template which will have it's own WWNN and typically 1 port on each fabric with a WWPN. From an FC perspective, typically the A fabric interconnect connects to your A SAN fabric, and the B connects to the B SAN fabric. Ethernet may be a port-channel to multiple upstream switches but if you don't have support for VSS, vPC, MLAG, etc. then you can just do "MAC pinning" which behaves a lot like a VMware vSwitch. This basically means each blade MAC address is going to be "pinned" to a specific logical uplink to your uptream network. With respect to blade redundancy you have a couple options. For FC you use traditional means like powerpath and for ethernet if your host OS doesn't have something built in you can enable fabric failover. This will move the vnics you create to the remaining fabric automatically if you lose a fabric interconnect. I generally don't do this for ESXi hosts but I'm going to use it for things like RHEL hosts. Fabric interconnects can connect with upstream (northbound) switches using LACP/802.3ad. Blades themselves however won't be able to do this. This generally isn't a problem because you're either going to run VMware on the blade which has built in mechanisms (load based teaming is simple and effective for example) or you have the fabric failover option I mentioned previously. Also you get to define your MAC pools, WWN pools and UUID pools yourself. This means you can pre-assign what you want blades to be and you can start stuffing metadata in your service profiles. This is tremendously helpful for pre-zoning, automated provisioning, etc. Also troubleshooting. You can absolutely get a UCS platform emulator. It's basically an API complete/management interface complete tool to get familiar. You can find that here: https://communities.cisco.com/docs/DOC-37827 and we frequently use it to test python and powershell code before exposing it to customers/our environment. If it works there then good odds it'll work on real kit. The only thing it won't do is boot a blade but you can start to understand the relationship between templates and policies and how the API is structured at least. Also I'm happy to answer any questions you may have on UCS so keep 'em coming. edit: for the TL/DR crowd: yeah it's basically magic.
|
|
#
¿
Nov 12, 2014 08:28
|
|
|
MC Fruit Stripe posted:I have like ten certs - they came with cards? Cisco provides wallet cards for CCNP and CCNA. Regarding verification, we verify all of our certifications since we need a certain number to maintain partner status with various vendors.
|
|
#
¿
Dec 1, 2014 06:25
|
|
|
Zero VGS posted:Counterpoint, switches can easily have redundancy too, and your switch/router configs had drat well better be in your existing backups. As a CCIE I would blow my brains out before I tried managing DHCP leases for servers, PXE, phones, desktops and wireless clients on a switch or router. Set up DHCP relay and leave that poo poo to server admins and make sure helpdesk monkeys can get read-only access through MMC or some web based management. Nevermind the whole "dynamic DNS updates" aspect of it. Windows 2012 makes it pretty easy and if you're the network guy it's nice to delegate poo poo like DHCP management to others. quote:I've also done a decade of Windows 2003 dhcp servers and holy hell why would you trust your network's connectivity to anything running Windows? We trust windows to provide authentication services for several thousand users every day which is pretty important for accessing network resources. I don't mean to dogpile here but Window is pretty reliable these days and even in 2003 if you're seeing frequent blue screens you should determine the root cause.
|
|
#
¿
Dec 29, 2014 09:51
|
|
|
Misogynist posted:Set up the simplest possible thing that doesn't create recurring maintenance work for you, like a hosted WordPress blog for people to post company news and events. If people use it, they will start coming to you with "hey, can we use the intranet to _____?" questions, and you can either find modules or start building requirements for the next iteration of the intranet. If they don't use it, you've done no more work than necessary. We're going to take exactly this route. We've got Confluence setup to house documentation, etc. but I don't want to try to hammer that into being our intranet as well.
|
|
#
¿
Dec 29, 2014 23:21
|
|
|
Zero VGS posted:There's a lot of dogpiling here when there's not even that much consensus. If you guys think best practice is letting Server 2012 handle it, that's cool, but I wouldn't trust it even if it were an option for me (which it isn't because my only servers are in Azure and I don't even want to imagine how that is supposed to work over VPN). The consensus is don't do this on a Cisco switch (or Juniper or Arista or whatever.) It's basically opening up a future world of pain and anyone that follows you into that environment will curse your name. Everything you mentioned about people going around processes and updating things or breaking things behind the scenes is equally true of dedicated network hardware (probably more likely since most environments I walk into don't bother with TACACS+ or even RADIUS to provide AAA for the devices. Just use admin and a shared password..) If you have concerns about people logging into your DHCP server and loving with things you lock it down and park it behind a firewall. Don't give it internet access and push updates to it only via WSUS and whatever policy you feel is appropriate. Most people don't seem to want to factor in processes and people into the uptime equation but it's probably the single most important piece of the pie. These are the sorts of network services best left to a dedicated server. I wouldn't enable the DNS server on my routers and point clients to that for example. Use your switch CPU cycles on something useful like routing protocol updates or maybe to provide a hardware NTP source for your virtualized DCs. The only time I would consider doing DHCP on the switches is if I have literally zero server footprint on-prem and my company is small enough/I'm the only guy. Even then I would consider buying a couple cheap boxes and running a local DC/CA, RADIUS for wireless auth, local NTP, a proxy server, and DHCP. At my primary employer we do exactly this and whenever we get a new printer or wireless access point or something they don't have to call me to setup the reservation. I'm drat sure not going to make regular changes to the single most important piece of my infrastructure or give the keys over to one of our junior IT dudes. quote:Okay, well, I might look into that. Let's say something wacky happens like oh, say, finance department lets the company credit card expire (like just happened, whups) and so Microsoft shuts down my Azure account without notification. Nobody can log into the network anymore because they lost access to domain controllers! This is a very real problem you're just going to have to make sure the finance department accounts for appropriately. I would see if Azure will offer you net 30 payment terms so your finance people just get a bill from microsoft to pay with whatever means (I dunno if they do this yet but it's possible!) Otherwise this will just end up being a pain point for the future.
|
|
#
¿
Dec 30, 2014 02:11
|
|
|
Tab8715 posted:Eh, I thought the DHCP-chat was quite engaging but if we want to move onto a different topic how about getting jobs way outside your current residence? It seems this is less of a problem in our industry but curious to see what's everyone else's take on it is. If you've got an in-demand skill set you can make this work. I left the southeast US to move to the San Francisco bay area in 2007 on nothing more than a VCP and a prayer. Expect to have a LOT of phone interviews and your first job in the new location is pretty likely to be a piece of poo poo and you're going to totally gently caress up your first house/apartment in the new location as well.
|
|
#
¿
Dec 30, 2014 05:01
|
|
|
Loten posted:Let me know if this is the wrong thread for this question, but I was not sure where it would fit. For vendor contacts make sure you include things like contract IDs, who your contract administrator is for it and things like circuit IDs, etc. Sometimes you'll go to open a support case with contract ID in hand and be told by the engineer that "So and so needs to open the case sorry!" so it's helpful to know who that is ahead of time and make sure you can open cases on behalf of that contract. This has bit me a couple times in the last few years.
|
|
#
¿
Jan 8, 2015 09:04
|
|
|
TWBalls posted:That's kind of my point. The company that you saved that million dollars for, what did they give you? Probably a pat on the back and an 'atta boy'. Well he took that accomplishment to justify his pay for a new job. Seems like a worthwhile payoff. If your company isn't financially recognizing your accomplishments then take what you've learned somewhere else until you feel like you're being fairly compensated.
|
|
#
¿
Jan 21, 2015 03:45
|
|
|
I moved out to the bay area to jump on some of the crazy stuff going on out here. I'd say it's worth throwing out an absurdly high number just to see if someone bites. The weather out here is pretty awesome and there are parts of the bay area where even 150k a year can get you a pretty decent place to live in a nice area. It won't be downtown San Francisco but if you've got any sort of family life you wouldn't want to live there anyway. Just ask for the moon and start from there.
|
|
#
¿
Mar 4, 2015 01:18
|
|
|
Bhodi posted:Downside: You actually have to pass the CCIE. I have an old co-worker who's slammed his face into that wall three times. That third failure has got to sting. Failures are pretty demoralizing because you don't find out for sometimes several days if you passed or failed! I took it the first time on a Friday expecting to have failed and when I got the notice at like 2am that I did it was a kick in the balls. The second time around I again took it on a Friday. I frantically mashed F5 on the status check page until it locked me out of the system for 48 hours. I was pretty sure I passed and I found out like 3 days later that I did. Those days were very very long.
|
|
#
¿
Apr 3, 2015 21:31
|
|
|
GreenNight posted:The company I work for won't pay their cut of your health insurance unless you get a physical every year. Married? Your wife has to get one too. Should probably be doing this anyway..
|
|
#
¿
May 1, 2015 02:46
|
|
|
AreWeDrunkYet posted:Once you find a job, is there any obligation to tell recruiters "thanks, but no thanks"? Will ignoring calls/e-mails have any impact next time you're looking for a new position and your name crosses their desk again? Unless you're a jerk about it I wouldn't worry about it too much. I sometimes linkedin connect with recruiters that work directly for companies I'm interested in just in case and I've had some hit me up every year to see if my situation's changed.
|
|
#
¿
Jun 12, 2015 20:09
|
|
|
Rancid (http://www.shrubbery.net/rancid/) or other. NX-OS includes egrep and supports multiple pipes... Drops the need for a 'do' prepend in configuration mode and depending on platform gives you access to some coreutils. I believe IOS-XR also includes vim among other things.
|
|
#
¿
Jun 14, 2015 20:15
|
|
|
adorai posted:You have to pay all your own payroll taxes as a 1099. It will work out to net the same, with all the headaches of being a business instead of an employee. You'll actually pay slightly more since your employer isn't covering a percentage of it. Also as a W2 you'll probably get your health insurance subsidized and of course vacation and sick time. My 1099 rate is actually ~160-200% of what my hourly rate would be as a W2 depending on how much I don't want to do the job.
|
|
#
¿
Jun 23, 2015 23:11
|
|
|
RFC2324 posted:I know plenty of people who have commutes like that here in the DFW area. One co-worker does 140 miles each way. The admin maintains and operates where the architect plans and builds to hand over to the sysadmin. Also I've known plenty of folks take a painful commute because it's pretty much the only option for a job or they just have to have the benefits.
|
|
#
¿
Jul 7, 2015 18:21
|
|
|
psydude posted:This is a pretty common misconception among a lot of the customers I work with. It's pretty rare that anyone I know at my company or the other companies in the area works more than 40 hours per week, and if we do then there's comp time involved (plus consulting firms tend to be incredibly flexible with works hours due to their nature). Occasionally some projects will see you spending several weeks out at another customer site across the country, but that's generally a once or twice a year thing, and to me it's generally worth not having to drive to the same office every day and deal with the same environment. Like all things, it depends on the individual - if you're married and have young kids, or are taking care of a sick family member or something, then yeah, it can be burdensome. But if you're single and like to travel, it's great. When I was traveling as a consultant I'd typically fly in Sunday night or Monday morning and fly home Thursday afternoons. I typically didn't spend 40 hour work weeks onsite unless a shitstorm was brewing or if they basically bought an expensive resident to do staff augmentation. It's definitely worth looking into if you're single and/or want to travel. Also depending on your region you might spend a lot of time at "home base' too.
|
|
#
¿
Aug 12, 2015 15:41
|
|
|
Internet Explorer posted:I have run into that exact thing more than once. Doing consulting / MSP type work really makes you hate your fellow IT workers. The worse offense I found of this was a company with ~250 employees and every one of them logged into their desktop with 'DOMAIN\Administrator' and the password was the name of the company. They had the worst IT department I'd ever seen. We eventually fired them all and fixed everything but it was pretty amazing.
|
|
#
¿
Aug 19, 2015 20:42
|
|
|
vibur posted:"I'm sorry but I'm not allowed to disclose that information." This is a good way to deflect that question. Say it's covered by an NDA.
|
|
#
¿
Sep 11, 2015 17:50
|
|
|
quote:If you want to manage Cisco, F5, EMC, etc, you probably want Puppet/Chef, since those are the established players with vendor support. Salt and Ansible are slowly coming, kind-of, but the Ansible cisco support is kind of hacky SSH stuff. Managing cisco with Puppet right now is a bit of a kludge with limited device support. You need something that supports nx-api (or the OnePK comedy option) and bash or 'guestshell' which pretty much means 9k/3k. My hope is that Cisco notices Ansible and develops some proper glue so I'm not limited to just 2 data center switch platforms. This is mostly for 2 reasons: 1. I don't want to have to deal with puppet proxy servers 2. You can't install an agent on catalyst switches, any router platform, firewalls, or the nexus 5k/7k platform I don't think EMC has any devices currently that support being managed by Puppet and F5 requires a proxy box to send commands to the device. edit: Here's another guy working on some cisco extensions: http://jedelman.com/home/ansible-for-networking/ it's just early for that kind of thing.
|
|
#
¿
Sep 17, 2015 04:50
|
|
|
Judge Schnoopy posted:I walked from a help desk manager promotion to a network admin offer, and the change can be pretty jarring. I went from closing 10-20 tickets a day to seeing one ticket a week. It felt like running full speed in to outer space; no matter where I looked there were no problems to fix and I was flailing for something to grab on to and fix. One of the first things I do in that position is document everything I have, how I'm monitoring it, etc. Then I start filling in the time between meetings and conference calls for improvements like setting all my devices to do AAA against AD, centralizing logging, roll out RANCID (or experiment with things like Oxidized), cleaning up the monitoring, etc. Basically sounds like you got a good handle on things.
|
|
#
¿
Sep 22, 2015 23:14
|
|
|
evol262 posted:Real talk - 99% odds they're not doing it either, and they think "sysadmin+config management" (or maybe sysadmin+scripting) means "DevOps environment!" Don't sweat it. I got into this stupid argument with a coworker the other day. He said something like "the moment you talk puppet we're in devops..." and it was a hell of a labor to explain that Puppet is configuration management not Devops and he wouldn't say that stupid poo poo if we were talking about VMware Configuration Manager.. I'm glad he's going to be gone soon. He keeps wearing me down with that type of poo poo.
|
|
#
¿
Oct 8, 2015 03:16
|
|
|
Kashuno posted:one of the big things my boss is looking at is statistics and replication on a per VM level, rather than a per volume level. From what I've read, it doesn't seem possible/easily doable on Tegile especially compared to Tintri. Does that sound accurate? We just deployed ~300TB of Tintri for a customer's ESXi environment. This is on top of ~150TB they already had and they seem to like them a lot. We had one deployed in our lab and I love it because I hate storage and find it the most boring thing ever. Tintri is great because you don't really need to worry about managing it and it can bombard you with all kinds of useful information.
|
|
#
¿
Nov 20, 2015 08:16
|
|
|
Kirios posted:stp sucks though and vss should be used instead where possible. We're doing more and more L3 everywhere. An IGP of some sort with BFD is pretty awesome. We use VXLAN whenever people need to vmotion a VM from one access switch to another.
|
|
#
¿
Jan 7, 2016 00:22
|
|
|
KillHour posted:l feel like that salary range is a bit low for someone with a CCIE or CCAr. Government jobs you do for the pension not the pay. In fact the salary range is probably based on years of government service and has nothing to do with actual experience.
|
|
#
¿
Jan 16, 2016 04:32
|
|
|
BaseballPCHiker posted:At what point do you say enough is enough when it comes to interviewing for a job? The most I've ever had was a phone screening, 1st interview that was informal, and then a second more technical interview and overview of future projects and IT needs. If I ever have to past that I think I'll just say no unless I'm really desperate. It really depends on the position I think. For example our post-sales people end up going through 3 phone interviews and the last one is a 4 hour face to face. The first phone interview is usually the recruiter screener asking things like "have you ever..." and "do you feel comfortable with..." Usually 30 minutes. The next interview is usually the hiring manager trying to just understand their personality/attitude how they approach poo poo. Another 30 minute to 1 hour call depending. After that there's a basic technical screen to sniff out resume bullshit. We try to keep those at ~30 minutes even when they're going well. Assuming thats all good we'll bring you to the office for a halfday session that may involve whiteboarding, presentation, etc. If you're traveling then we'll put you up in a hotel and pay for airfair/car. If it's a sales or pre-sales then all that poo poo goes out the window and you end up talking to a couple of the account managers, the EVP who owns sales engineering, presales engineering, the EVP who owns professional services, and at least one of the owners.
|
|
#
¿
Feb 2, 2016 22:02
|
|
|
MrMoo posted:Commercial FreeNAS, the software lives on a USB stick and you provide the hardware. There are quite a few players in this space, including Open-e, Wasabi Systems, OpenFiler, Rockstor, ... None of these are remotely close to what Nutanix is... Nutanix is basically a storage/server system rolled up into a box where the local storage in every node gets presented back to all of the nodes as one big shared volume. Usually used for running some form of virtualization. A close analog would be vSphere + VSAN but nutanix also provides a lot of helpful tools to get up and running through wizards. It'll go through and setup the shared storage and instal your hypervisor of choice while you go to lunch. edit: Someone just posted this helpful link in the virtualization thread http://nutanixbible.com/ (thanks TwoDeer!) 1000101 fucked around with this message at 03:26 on Feb 3, 2016 |
|
#
¿
Feb 3, 2016 03:24
|
|
|
Dark Helmut posted:I feel for Zima for a couple weeks in the 90s. NEVER AGAIN. Pretty popular here in the bay area. I'll have on occasionally but I have them replace the vodka with rye.
|
|
#
¿
Feb 5, 2016 18:18
|
|
|
Foe Hammer posted:I have a fairly good background in network security but I have gotten to the point now where I feel like I have to learn C++/ Python to further myself. Now I have self taught myself by downloading various training movies via torrent and have even used http://www.professormesser.com/ I would pick one language and focus on that or you'll go mad. I'd say start with Python as you'll find it helpful to pull data from devices to generate reports, etc. I found the best way to actually learn it is to actually have some kind of simple projects. When I was learning python I basically went through some basic stuff at Coding Academy and a few youtube videos. I then went right to work on writing a script to pull a list of hostnames and IPs from a system and output a CSV via it's REST API. I went from that to building a little tool to provision virtual machines out of VMware's vRealize catalog via their REST API, then generating reports from different areas, etc. You'll find yourself on Stackexchange an awful lot but that's probably fine as long as you try to understand what the code you're shamelessly stealing is doing. After a few months of doing that I'm pretty comfortable and finding myself googling less. I can also mostly follow other folks code. I always felt that without something to apply it on, trying to learn basic programming would be nigh impossible.
|
|
#
¿
Mar 1, 2016 18:34
|
|
|
RFC2324 posted:How is googling it not learning it? I don't mean just Google up an answer without thinking about it, but I'd bet you can find all the info you need to begin formulating a plan in an hour or 2 of research, and then the details specific to your org to get everything set up. Usually people google things within the context of whatever it is they're trying to do and often they end up searching for the wrong solution. Then you end up with poo poo like an /etc/shadow and /etc/passwd file being distributed to a couple dozen servers from git versus just using LDAP because you haven't actually learned anything but how to do a task which may not even be the right task. A DR plan is a perfect example where someone might google for ways to failover all of the servers to a remote location and get all the backups in order and do everything right with the servers. Then come to find out that none of the network plumbing is there, nobody bothered to come up with alternative workspaces, zero planning was done for communications/telco, nobody could figure out how to find anyone when poo poo hit the fan or a myriad of other things. Though you could have someone get lucky and google 'how to dr plan' and get a couple decent starting points. Maybe I'm just tired of people depending on google to do their thinking for them.
|
|
#
¿
Mar 23, 2016 19:54
|
|
|
DigitalMocking posted:I'm not a fan of certs myself. Paper tigers are the worst kinds of teammates to get stuck with. A certification isn't any different than a line on a resume. Just helps qualify someone through a basic screening process and set an expectation as to what skill level may be required to do the job for the would-be candidate.
|
|
#
¿
Apr 5, 2016 03:59
|
|
|
Fiendish Dr. Wu posted:
You're hired!
|
|
#
¿
Apr 8, 2016 03:40
|
|
|
The worst resume I read was 4 pages with an 8 point font. It also looked like 3 different people wrote it from the perfectly cogent 4 paragraph intro to the broken english bullet points for the first half of their work experience to the even less cogent second half. Hiring is the worst thing I've had to do.
|
|
#
¿
Apr 8, 2016 05:04
|
|
|
H110Hawk posted:Datacenter porn? We have 4 new colo cages going in simultaneously. Here are some build progress pics: Is that 3rd QFX just a spare in case one of the other switches dies? Also, are you doing virtual chassis?
|
|
#
¿
Apr 9, 2016 20:22
|
|
|
jaegerx posted:Sounds like cissp role. Private sector pay is largely irrelevant in most public sector jobs. They tend to set the pay scale based on years of service and the work itself is pretty slow paced. People take the job though because of retirement benefits and so they can follow basic procedures from 9-5 every day.
|
|
#
¿
Apr 10, 2016 07:15
|
|
|
MC Fruit Stripe posted:We've got this new manager over in development who, any time there is a delay on a request (because things happen and people get busy), will ask what the SLA is on these kinds of requests. Some larger orgs I've worked at have actually had SLOs (like an SLA but an objective instead of an agreement) defined for internal "customers." For example an internal DNS record might take 5 business days, a global record 10 business days. Vulture Culture's pretty much spot on though. Most orgs you just need to set an expectation for requests. "It takes 24 hours to deliver you a web server because we have to build the VM, install software, create a DNS record, get an IP, etc. All while still doing our day job putting out fires!"
|
|
#
¿
May 13, 2016 20:37
|
|
|
|
| # ¿ Apr 29, 2024 12:27 |
|
|
It's risk management. There are values to using snapshots as a part of an overall backup strategy as long as it's not your backup strategy. Most people saying "snaps aren't backup" are generally conveying the point that if you're only doing snaps then you'll probably run into a number of unrecoverable scenarios. Some involve array failures, some involve security breaches, some involve human error. I have plenty of customers that have pretty tight needs and they still depend on external devices as the "oh poo poo!" catchall. Snaps are just the first line of defense for them. A backup is a backup until it's not there anymore so you try to have a couple layers of backups. I'd rather go back to last week's financials than start trying to re-assemble from spreadsheets scattered over dozens of laptops.
|
|
#
¿
May 18, 2016 03:07
|
|