Useless Conversation

[Murgos]

Needless to say, as the FCC is predicting the end of the land line in about 6 more years, this has telemarketers and collection agencies running scared.

Your article is quoted from Avaya.

I picked that one as it had discussion on the contents of the FCC's report, which was linked at the end of the article:

So when is the existing PSTN network going away? According to this group, sometime before the next decade.
You can read the report yourself here: http://transition.fcc.gov/oet/tac/TACJune2011mtgfullpresentation.pdf

You guys, and Avaya are reading that wrong.

The circuit switched voice portions, i.e. the analog call routing infrastructure is probably going away.  The PTSN will still be there and so will the twisted pair land line it will just have all digital routing, which it mostly does already.

[Yegolev]

I suppose they must be talking about the support for rotary phones.  But maybe that's already gone, I don't know.

[Minvaren]

Not 100% sure there, Murgos.  The report says:

1.Target 2018 as the end of the PSTN.
2.Develop timeline  to ensure smooth transition which addresses stranded assets
3.Assure that mobile and/or broadband replacements are available everywhere PSTN 
is currently provided. The need will be greatest in rural areas.
4.Update the National Broadband Plan to support the PSTN transition.

PSTN replaced by broadband or mobile seems to indicate that POTS is going away.  T1s and Data circuits will still likely exist, as you mention that these are already packet-switched anyway.

[Murgos]

Every year you can move more and more data across POTS and it's everywhere.  Saying things like "Broadband" is not indicative of the medium used to carry it.

[Cheddar]

The blog is on Avaya.  You know, a major IP voice player.  That was my point.   

[Merusk]

Whatever, Dinosaur-lad! 

[Khaldun]

My meeting was almost as bad as I thought, but not quite. There were actually a few decent ideas about how to get on to the next step, which was pleasing and surprising.

[Ironwood]

The new Amazon Storage Gateway is actually quite cool.

[Soln]

Indeed.

[Yegolev]

What sort of data speed do you get?

[Ironwood]

Not done any massive data push testings yet, but the actual snapshots themselves seem to take a lot less time than you'd think.

Honestly, if I was going to gush about ANYTHING on these boards, it would be the leaps and bounds that Amazon are doing with Cloud and pushing the boundaries of consumable IT.  They've pretty much just given me a Cloud backup system for my clients right when I needed one and it took, no joke here, 30 minutes to setup and get working.

AWS is shockingly good.


I'll let you know when I start pushing servers out of it.

[Yegolev]

My experience is with EMC arrays, the latest ones able to take snapshots right away with background sync, but I really don't know how that would work if AWS doesn't already have a copy of my data.  I suppose there is some local processing to get a timestamp and then begin collecting changes...?

[Ironwood]

The way it works at the moment is everything that's put on the Gateway syncs up to AWS and then Snapshots at a time of your choosing.

So, you could stick a bunch of data on there, start snapshotting every night at 9 and hey presto, online backup.  You could even be quite clever and put the data as a DFS partner and that way you're not only covered locally, but also out there.

You can mount those snapshots on EC2, so in case of disaster, you can be up and running again in five minutes. 

I suspect that moving legacy stuff ON to this could be time consuming, but for smaller organisations and for startups, it makes sense to me.  I've been snapshotting since setup and buggering about and so far I owe.... 1 cent.

Which is nice.

[Ironwood]

  I suppose there is some local processing to get a timestamp and then begin collecting changes...?

Oh, and Yes.  There is.

[MuffinMan]

Someone erroneously sent an e-mail to the entire company this morning. We're now going on an hour of everyone using reply all to say "I'M NOT TONY, WHY DID I GET THIS E-MAIL?"

[RhyssaFireheart]

Someone erroneously sent an e-mail to the entire company this morning. We're now going on an hour of everyone using reply all to say "I'M NOT TONY, WHY DID I GET THIS E-MAIL?"

Oh dear gods.  I hated when that happened at the last company.  Or the folks replying that they needed to be taken off the particular distribution list used.  IT actually had to go in once and remove the emails from folks' inboxes to stop the reply all spam, it was so bad.

[ghost]

That makes me smile a little.  Technology is wonderful.   

[Ironwood]

It is.

There's a powershell command to wheech the offending e-mail from the store.  I reccomend bookmarking it.

[MisterNoisy]

Oh dear gods.  I hated when that happened at the last company.  Or the folks replying that they needed to be taken off the particular distribution list used.  IT actually had to go in once and remove the emails from folks' inboxes to stop the reply all spam, it was so bad.

There needs to be some sort of way to disable the 'Reply All' function for repeat offenders.  There's people here that literally use it in response to every single message they receive.  Every one of them is a top-tier assgoblin.

[Khaldun]

Literally it's about three days here in between someone replying to the faculty-staff digest as if they were replying to a person who had a message in the digest. Always at least a bit embarrassing, occasionally spectacularly so. I think the one about the person who said they'd be by someone else's office to "give them something special they've been wanting for a while" was especially

It doesn't matter how many times it happens or how big and all-caps the warning at the top of the digest is: there's always fresh blood ready to do it again.

[Yegolev]

You can mount those snapshots on EC2, so in case of disaster, you can be up and running again in five minutes.

Do you do this with databases?  Because I don't care too much about other data, it's relatively small, granular, and easy to manage.

EMC says we can snapshot a running Oracle DB and mount the snapshot elsewhere, but so far I haven't had that work.  This could be due to my lack of competent DBAs, though.

[Salamok]

Literally it's about three days here in between someone replying to the faculty-staff digest as if they were replying to a person who had a message in the digest. Always at least a bit embarrassing, occasionally spectacularly so. I think the one about the person who said they'd be by someone else's office to "give them something special they've been wanting for a while" was especially

It doesn't matter how many times it happens or how big and all-caps the warning at the top of the digest is: there's always fresh blood ready to do it again.

BCC is your friend.

[Cheddar]

I love people with 1+ page signatures.   

[Ironwood]

You can mount those snapshots on EC2, so in case of disaster, you can be up and running again in five minutes.

Do you do this with databases?  Because I don't care too much about other data, it's relatively small, granular, and easy to manage.

EMC says we can snapshot a running Oracle DB and mount the snapshot elsewhere, but so far I haven't had that work.  This could be due to my lack of competent DBAs, though.

I have done it with SQL chaps, yeah, but really small scale.  No idea how it would work for larger databases.  If I was going to do that I'd probably just use the AWS Database stuff anyway.

Why didn't it work with EMC ?  Surely a snapshot is a snapshot ?

Edited to Add :  I should make it clear right now I know FUCK ALL about Oracle.  There are cockroaches with more experience in that area.

[Ingmar]

Those cockroaches get like $250/hour as consultants.

[Ironwood]

Sure, but they have to work with Oracle.

Karma.

[Trippy]

Snapshotting a running RDBMS is tricky especially if you aren't blocking writes temporarily.

With regards to Amazon if you are on EC2 instances with an EBS volume, Amazon's EBS snapshot functionality does support XFS's freeze feature. So you can block writes to the DB temporarily, flush everything to disk, freeze the XFS volume the DB files are on, start the EBS snapshot, unfreeze the volume and remove the global write lock. The snapshot will continue on it's own using the frozen image of the volume.

[Salamok]

I was under the impression that with some DBs a snapshot is the only way to back them up reliably.  For example if it is a transaction based DB with many foreign key constraints, attempting to dump the data via SQL isn't fast enough to maintain the integrity of the database.

[Ironwood]

Snapshotting a running RDBMS is tricky especially if you aren't blocking writes temporarily.

With regards to Amazon if you are on EC2 instances with an EBS volume, Amazon's EBS snapshot functionality does support XFS's freeze feature. So you can block writes to the DB temporarily, flush everything to disk, freeze the XFS volume the DB files are on, start the EBS snapshot, unfreeze the volume and remove the global write lock. The snapshot will continue on it's own using the frozen image of the volume.

Which is fair enough, but I do it with small ones that aren't getting hit much and the snapshot is at 9 at night and who isn't at the pub then ?

 

[bhodi]

You still need some way to quiesce the database before you take the snapshot. If you're running oracle 9i+, you can look here or just google 'oracle quiesce snapshot' for more info. THEORETICALLY it will work, but I've never had to try.

[ghost]

Can we get a data/database management and implementation subforum or thread?   

[Sky]

I love people with 1+ page signatures.  

I do my part by not signing email.

To re-rail the thread: I found a supply of honeycrisps. In friggin' JANUARY.

[HaemishM]

I'm still getting honeycrisps in the stores down here. Had one for lunch actually. It really is nectar of the Gods.

[Yegolev]

Can we get a data/database management and implementation subforum or thread?   

We had to read all your clean-teeth crap.

[Yegolev]

EMC is supposed to handle the write sync on the frame, and that might be awesome or it might not, but Oracle detects a crash and demands some recovery.  Which might be awesome or terrible, depending on what is going on.  Usually a recovery is fine, but we have those cockroaches that Ironwood spoke of, and I'm not allowed to exercise my meager Oracle knowledge due to not being paid to do DBA work.  So you tell some people that you have a crash-consistent snapshot of the DB, and they nod, but then Oracle starts sending messages to them at the sqlplus prompt and they freak out.  Because those messages aren't in the For Dummies book, apparently, so I must have broken it with my storage-device-based black magic.