#arpnetworks 2017-05-05,Fri

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
***tabthorpe has quit IRC (Ping timeout: 258 seconds) [00:49]
......................................................... (idle for 4h42mn)
Nahual has joined #arpnetworks [05:31]
..... (idle for 24mn)
nitemare has joined #arpnetworks [05:55]
qbit1 has joined #arpnetworks
trobotham has quit IRC (*.net *.split)
qbit has quit IRC (*.net *.split)
mercutio has quit IRC (*.net *.split)
nitemare is now known as trobotham
[06:02]
qbit1 is now known as qbit
mercutio has joined #arpnetworks
orwell.freenode.net sets mode: +o mercutio
[06:11]
mercutio has quit IRC (*.net *.split)
mercutio has joined #arpnetworks
ChanServ sets mode: +o mercutio
tabthorpe has joined #arpnetworks
tabthorpe has quit IRC (Changing host)
tabthorpe has joined #arpnetworks
[06:26]
....................................... (idle for 3h10mn)
Nahual has quit IRC (Ping timeout: 240 seconds)
Nahual has joined #arpnetworks
[09:37]
........ (idle for 37mn)
ziyourenxiang has joined #arpnetworks
ziyourenxiang has quit IRC (Changing host)
ziyourenxiang has joined #arpnetworks
ziyourenxiang has quit IRC (Client Quit)
ziyourenxiang has joined #arpnetworks
[10:16]
sjacksoGotta be careful with an offer like that mike-burns. If you were sufficiently bloody-minded you *could* implement microsoft word in three lines of C. [10:22]
........... (idle for 51mn)
***ziyourenxiang has quit IRC (Quit: Leaving) [11:13]
.................................................................... (idle for 5h37mn)
caseyandginaI don't really understand the difference between arp VPS and arp thunder...I expected the thunder server, which has 4x the RAM, SSD, and Sandy Bridge over Nehalem, to be faster for the same operations. But it's the opposite.
not quite sure what I expected, but I didn't expect the hardware to look identical to an arp vps on the thunder, though that's certainly convenient.
but then there's the obvious question of why does it sound like I'm getting a lot more for the same money with a thunder than a vps?
well, I'm not really doing a full apples-to-apples comparison yet, kernel, compiler, etc. versions are not yet in sync
nonetheless would love to hear more insights
[16:50]
HAS_A_BANANAHuh that does sound unusual. Thunder is intended to give dedicated (or better!) performance, particularly with disk IO. I haven't tried Thunder yet so I can't speak first-hand, but I am tentatively planning on migrating my company's Metal to Thunder. Your experience gives me some pause.
(Not that my migration would be instantaneous anyways - it would be setup, benchmark, etc before cutting-over)
[17:05]
mercutiocasey: what is going worse in particular? [17:06]
perlgodresurrecting a topic from above, i also ordered a second VPS and have no IP for it
put in a ticket early this morning, still waiting on a reply
boots up fine, just no networking connectivity
[17:10]
caseyandginamercutio: compile time for gcc, etc.
perlgod: do you have an IP block already?
[17:18]
mercutiocasey: that's interesting, maybe send in a ticket? [17:19]
perlgodyeah, just a single IPv4 address for a metal server i also have [17:19]
mercutiocompile time is usually cpu related [17:19]
caseyandginamercutio: well like I said I'm still on inconsistent versions of stuff, so I'll get things in line first. [17:19]
mercutioah,
i have found clang is faster than gcc in the past
oh you were using gentoo
[17:20]
caseyandginabut what *is* thunder? Just the same as VPS on different hardware? [17:20]
mercutioit's dedicated resources, so it should be more consistent performance wise
having slower cpu seems odd
[17:20]
caseyandginalogically the same software managing things? just a different config/pools of hardware underneath?
and I guess no overprovisioning of resources
[17:21]
mercutioyeah and no over provisioning of cpu cores.
err over subscribing
[17:21]
caseyandginawhat about RAM, etc.? [17:21]
mercutioram is as it is for the plan
oh ram is never over subscribed
that'd suck :)
[17:22]
caseyandginamy first experience with VPS's was at vpslink.com, where we seriously overprovisioned everything to the max. Anyone who actually put any significant load on their VPS got reprovisioned onto a
"bad customers" server with all the other people who actually used resources
so they ended up having the most awful experience while the people who didn't actually do much got great performance.
[17:22]
mercutioheh [17:23]
caseyandginasince, most money was being paid into the business by the latter group. pretty lousy business model, IMHO. [17:23]
mercutiodisk performance on everything is more consistent than it used to be
with the usage of shared storage.
and ime sluggish disk performance is one of the most frustrating issues to have with a vps
[17:23]
caseyandginaI haven't kept up, but inconsistent I/O has always been my biggest issue with VM's at places I've worked. [17:24]
mercutioyeah, well shared storage makes it a lot better, as the load is spread out a lot more [17:24]
caseyandginalast company had our database servers on self-managed vmware hardware - as a DBA that certainly made things "interesting"... :/ [17:25]
mercutioso you'd need more heavy users at once to notice [17:25]
caseyandginathey ended up provisioning dedicated storage when they switched to a new database platform, and thus no longer had a need for me, lol
I'll have to do some more comparisons in detail while I still have both services going.
but the Thunders seem like a much better deal than the regular VPS's, wondering what the catch is ;)
[17:25]
mercutiowell the beginning prices start a little higher
but yeh i reckon the price point is good
[17:27]
caseyandginaHAS_A_BANANA: personally, I feel that while Thunder is a VERY compelling service offering for many, if your company is able to manage dedicated servers without issue then I'd stick with that. The advantage of Thunder is removing some administration overhead of getting servers all set up and configured for redundancy, etc.
with convenience comes some loss of flexibility/control
[17:29]
mercutio: ah, well I went from a $60/month VPS to a $90/monnth Thunder. For that 50% increase in cost I went from 4GB to 16GB RAM, 128GB to 400GB SATA, addition of 160GB SDD for 128 --> 560GB total storage, a newer/faster CPU generation, 6 --> 10TB bandwidth (don't care honestly)...
and it should have been $63/month as I had 4x CPU cores configured on the VPS
[17:35]
***ziyourenxiang has joined #arpnetworks [17:36]
mercutiocool
16gb of ram for $90 hmm
that does seem good
[17:36]
caseyandginathe storage was nice too - now I can dump all my personal backups there too.
just once in a while, primary is a portable ssd
are there any hardware tiers behind the scenes? how is older hardware phased out?
one thing I was hoping for with this transition was to get better compilation times, mainly because building the Go application where I work and running unit tests on it takes about twice as long on the ARP VPS as my MacBook Pro
[17:37]
mercutioi wonder if the versions are doing something with that [17:41]
caseyandginathe mac is great hardware, but I'd expect a highly-tuned and more modern linux system to be faster than Mac OS bloat
for our app it's easy enough to be consistent - we target a specific Go and PostgreSQL version.
let me just get the O/S consistent as a foundation first ;)
I'm still running off the boot CD kernel, and ARP's version of the Gentoo boot CD is from 2013!!
up_the_irons: could you add this to the dropdown selector, and/or allow a text input for pasting a custom URL to an ISO? http://distfiles.gentoo.org/releases/amd64/autobuilds/20170504/install-amd64-minimal-20170504.iso
[17:43]
mercutioi'd expect modern linux to be slower than old mac system with the same cpu
gcc keeps getting slower and slower
[17:47]
BryceBotThat's what she said!! [17:47]
caseyandginalolol, kernel 3.10.7 on the Thunder vs 4.11.0 on the VPS. :/ [17:47]
mercutiokernel won't make much diff for compiling
gcc version is more likely to make a diff
[17:48]
caseyandginayeah, I was comparing compiling the same version of GCC, but using different versions of GCC to do that compilation :|
GCC whatever-it-was also probably being 4 years out of date.
[17:48]
mercutioyeah that's probably not a good tet
test
[17:49]
caseyandginaas for Golang, newer versions compile faster, but I'm not sure if they have optimized more heavily for Mac [17:49]
I should really throw them for a loop and put it on OpenBSD ;) [17:54]
mercutioopenbsd may not work on arp thunder by default
may or may not work
there was a bug in openbsd that required a workaround before.
but it's meant to be fixed in 6.1
[17:59]
........................................ (idle for 3h18mn)
***Nahual has quit IRC (Quit: Leaving.) [21:17]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)