Verizon Corporate Email - Android ?

[PSkeptic]

Including multiple locations, I assume, as weather and power related issues could bring down a single locations for more than 5 minutes beyond your control?

Yes. I work for a mutli-national corporation. One of the largest employers in France, actually.

A single location is not the email system. A single location is not the network.

Depending on site availability SLA's, we have gensets and backup links. Sometimes gensets to backup the gensets, and sometimes backup links for the backup links.

Our data center has 3 tiers of data backbones coming in. In fact, our telco pays us for routing some of their traffic.

I find that incredibly hard to believe.

Of course you do. I'm assuming you are not in a position to actually see systems implemented in an enterprise-size environment.

Well, then you would, in fact, be wrong about that. Granted, you can tell that I've never done an RFP regarding an email server, and on that you would be very very right.

I'm seeing you've never done and RFP for most anything enterprise grade. 5 9's is not an uncommon requirement for an enterprise. Generally speaking, there are stiff penalties for violating your SLA.

It's a rather easy deal, it's just not quick. As long as you can agree on a format to put the data in, and a way to segment the data between tapes, it's fairly easy to store that data, and have it sent back from their facility... it's just extremely inconvenient.

It's not easy. It's not quick. It's not cheap. Formats alone are a huge deal. Have you ever backed up a petabyte of data? Do you know how long it would take to generate that backup set? Do you know what the requirements for an enterprise email system is?

Seriously, imagine this: You decide to back out of Google Apps. So, now, you have to freeze all mail routing, get the tapes/disk sets/etc sent to you (Same day, if you want to pay for it, shipping costs alone for a multi-petabyte backup set would be enormous). Then, you have the time for restore to the new system (If that's possible).

So, you are looking at a full day outage for email. A single day without email can easily cost 40 million.

Any change is involves unknown. That's like saying that shareholders don't like you doing anything that you haven't already done before a thousand times.

Shareholders generally don't want IT to do things that haven't been proven. That's why the enterprise environments tend to like having minimal numbers of models of computers, like having a unified set of server platforms, generally have 3 environments (Development, QA, and Production), etc.

You are quite naive when it comes to how IT works in enterprise environments.

 

[byteware]

Yes. I work for a mutli-national corporation. One of the largest employers in France, actually.

A single location is not the email system. A single location is not the network.

That would have to be true.

Our data center has 3 tiers of data backbones coming in. In fact, our telco pays us for routing some of their traffic.

To be able to pay for that... drool. Us Public Sector grunts don't get such niceties... I have a few locations I'm still trying to get off of VPN access and onto a T1... when they find the money.

Of course you do. I'm assuming you are not in a position to actually see systems implemented in an enterprise-size environment.

Actually, you would assume wrong. What you CAN assume is that I'm not in a position to see it done with all the assurances that you are able to.

I'm seeing you've never done and RFP for most anything enterprise grade. 5 9's is not an uncommon requirement for an enterprise. Generally speaking, there are stiff penalties for violating your SLA.

3 9's, never 5 9's.

It's not easy. It's not quick. It's not cheap. Formats alone are a huge deal. Have you ever backed up a petabyte of data? Do you know how long it would take to generate that backup set? Do you know what the requirements for an enterprise email system is?

Email formats are easy. Choose the one that you can import into the new system.

As far as changing over, I would assume you would schedule a changeover, instead of doing it overnight? You would send everything on the servers a month before the changeover (not overnighted). Start working on restoring them into the new system. One week before changeover, send all the new data, and have it restored (you should have the restoration process down by now). Bring the gmail servers down, and your servers up for production, and overnight the last bit of data. There are a few hours that those emails won't be available (and that's not great, I understand), but this really isn't that difficult to figure out. The specifics might be trying (especially the import), but beyond that, this isn't rocket science.

And I can almost guarantee you that the vendor will be more than willing to help with the import.

So, you are looking at a full day outage for email. A single day without email can easily cost 40 million.

See my above scenario.

Shareholders generally don't want IT to do things that haven't been proven. That's why the enterprise environments tend to like having minimal numbers of models of computers, like having a unified set of server platforms, generally have 3 environments (Development, QA, and Production), etc.

Yes, very familiar.

 

[Dakota - Tom]

...Us Public Sector grunts...

Oh, Lord help us... the government is putting its email in the cloud... Wikileaks will soon seem like such a minor security breach ...

 

[PSkeptic]

That would have to be true.



To be able to pay for that... drool. Us Public Sector grunts don't get such niceties... I have a few locations I'm still trying to get off of VPN access and onto a T1... when they find the money.

Depends on which part of the public sector, actually. You would have no such issues working for the Dept of Education, or IRS. Even Treasury (They require 5 9 SLA's for SP's).

Actually, you would assume wrong. What you CAN assume is that I'm not in a position to see it done with all the assurances that you are able to.

Again, 5 9's is not an uncommon SLA for an enterprise. I can only guess you work in a moderately sized shop, not a multi-national enterprise or national (Multi-national, or even national levels are considered "Enterprise". Otherwise, it's consider Midsized, per IBM sizing specs).

3 9's, never 5 9's.

5 9's is quite common: 5 9's uptime SLA - Google Search

Email formats are easy. Choose the one that you can import into the new system.

Pray tell, what format can one use to bulk import into Exchange? How does one handle exceptions (Such as over-limits, etc)?

As far as changing over, I would assume you would schedule a changeover, instead of doing it overnight? You would send everything on the servers a month before the changeover (not overnighted). Start working on restoring them into the new system. One week before changeover, send all the new data, and have it restored (you should have the restoration process down by now). Bring the gmail servers down, and your servers up for production, and overnight the last bit of data. There are a few hours that those emails won't be available (and that's not great, I understand), but this really isn't that difficult to figure out. The specifics might be trying (especially the import), but beyond that, this isn't rocket science.

How do you propose we get end-users to NOT save anything in their inbox/calendar for 1 month? You would still have to overnight 1 month's worth of data. Or, one week's, or whatever your period is for.

The plain logistics of getting your data out of the cloud, and into your house is a giant exercise. One many companies who dove in early in the cloud computing craze learned the hard way.

Here's a great article targeted at CIO's in regards to "cloud computing":
The Case Against Cloud Computing, Part One CIO.com

One point, SLA's, get an entire section devoted to it.

And I can almost guarantee you that the vendor will be more than willing to help with the import.

Yep, any vendor will help you get your data in. Hardly any vendor will work with you to get your data out in an easy fashion.

See my above scenario.

And your above scenario still works in a day of downtime. Not feasible in a high-availability environment.

Yes, very familiar.

Then, you should know all about how to structure a HA system, and the inherent downfalls of giant data migrations, and why they can't be done in a day, as you propose.

 

[byteware]

Depends on which part of the public sector, actually. You would have no such issues working for the Dept of Education, or IRS. Even Treasury (They require 5 9 SLA's for SP's).

By US... I meant "us", not U.S. I'm a public sector Grunt... not federal, state. We aren't allowed to print our own money to spend. We have to only spend what we actually have.

Pray tell, what format can one use to bulk import into Exchange? How does one handle exceptions (Such as over-limits, etc)?

Microsoft can answer those questions for you, as they did for us when we brought everything from Notes to Exchange.

How do you propose we get end-users to NOT save anything in their inbox/calendar for 1 month? You would still have to overnight 1 month's worth of data. Or, one week's, or whatever your period is for.

I don't. I expect 1 Month out to be able to transfer the majority of the data.

The 1 week out to be have the process down, and get what they've saved since then.

Then on the last day, you get the rest of their data.

That last weeks worth of data should be able to be sent electronically, and therefore not result in much (if any) downtime.

The plain logistics of getting your data out of the cloud, and into your house is a giant exercise. One many companies who dove in early in the cloud computing craze learned the hard way.

Like I said, it's not easy. It's not cheap.

Yep, any vendor will help you get your data in. Hardly any vendor will work with you to get your data out in an easy fashion.

And your above scenario still works in a day of downtime. Not feasible in a high-availability environment.

Then, you should know all about how to structure a HA system, and the inherent downfalls of giant data migrations, and why they can't be done in a day, as you propose.

I didn't propose over a day, I proposed over a month.

I proposed transferring data in three stages. Bulk transfer a month out to give yourselves time to ensure the process works flawlessly. Another bulk transfer 1 week out to get the "new" stuff, and one last transfer on the day of.

 

[PSkeptic]

I'm done here. Come back when you've actually had some enterprise experience as a Systems Architect.

 

[Lars]

T1s are pretty 80's. It always amazes me that the phone companies are still trying to push 30 year old tech as cutting edge. It's the best they can do over twisted pair. Even a cheap $30 cable connection can do 50mbps dl and 2mbps ul which is better than a T1 line. Buy a real connection from them and you have slow LAN network speeds across a WAN in the form of 10mbps synchronous.

 

[byteware]

T1s are pretty 80's. It always amazes me that the phone companies are still trying to push 30 year old tech as cutting edge. It's the best they can do over twisted pair. Even a cheap $30 cable connection can do 50mbps dl and 2mbps ul which is better than a T1 line. Buy a real connection from them and you have slow LAN network speeds across a WAN in the form of 10mbps synchronous.

Yes, but we need a synchronous connection, and unfortunately, cannot use an asynchronous connection.

 

[Lars]

Im curious why? Ignore the 48 mbps of dl and you have a 2 mbps channel in both directions.

 

[Lars]

Cable co offers sych connections too. Just say'n might be worth looking at. Static ip too.

 

[byteware]

It's already been looked into, and what they offer in the areas of our remote offices isn't sufficient. I don't know whether or not they specifically offer other services elsewhere.