Microsoft exchange's future?

sizeon

Do you guys think on premise exchange is going to die out? With hosted exchange and office 365, i don't see why someone would want an inhouse exchange.

N2IT

Last two companies I worked for went to hosted. The current company I am with has been great. Very rarely does exchange suffer problems. My impression so far is a good one.

Chivalry1

To be honest I don't think there is a threat of it going away anytime soon. However the introduction of "Daas" [Desktop as a Service] cloud services may push more companies completely to the cloud. And it would only makes sense that if your desktop is hosted at a certain Cloud Company to also relocate the Email/Directory/ERP/CRM services there as well. I think DaaS is the pivotal point and a IT game changer.

But ultimately I believe all IT Professionals will be working for ISP and Cloud-Based service companies as subject matter experts.

N2IT

Nice post Chiv - I agree with DaaS being a significant game changer.

My current company has all but one server hosted along with several software, exchange etc. I haven't missed a beat to be honest. In fact we are finally transitoning to office 365.

sizeon

I disagree on your view on Daas. If your network is down then you don't have access to your machine. If your ISP is having latency issues, your desktop experience is going to be bad. The cons outweighs the pros imo.

cyberguypr

Most cloud-centric conversations here end the same way: yes, adoption will most likely grow. No, it's not going away any time soon.

Chivalry1

sizeon wrote: »

I disagree on your view on Daas. If your network is down then you don't have access to your machine. If your ISP is having latency issues, your desktop experience is going to be bad. The cons outweighs the pros imo.

Understood Sizeon....just providing insight from my point of view as a consultant. With the cheap cost of ISP data couriers lines, which most companies already have, the risk of a network outage/slowness are mitigated. I know one client company that literally has 4 data lines to ensure business continuity. I know for some small companies in remote areas this won't be possible, but these days you can get a extra data line from a local ISP for less than $100 a month.

dbrink

The cost of hosted Exchange gets very expensive the larger the organization is. For instance, 20k users at $3-4 per mailbox is $60-80k per month. You can easily build an exchange deployment for $400k that will last you a good 3-5 years.

jaz0nj4ckal

dbrink wrote: »

The cost of hosted Exchange gets very expensive the larger the organization is. For instance, 20k users at $3-4 per mailbox is $60-80k per month. You can easily build an exchange deployment for $400k that will last you a good 3-5 years.

I have always wanted to do the math on large environments; however, I have only been in 300> members. I am looking for hosted email solutions currently. Can anyone give any good recommendations on service providers? Coming from managing Exchange – are all he features present with most hosted Exchange services? Do I have availability for logging and message tracking as I can on the backend with Exchange?

alan2308

jaz0nj4ckal, I just happen to work for a company who offers hosted Exchange, so I can recommend a company.

DevilWAH

I do think Office apps will move more to the cloud as they are a prime candidate for it. Tried and tested and can be easily scaled to fit. In terms of very large organisations you stop paying a price per mail box and just end up as a set cost. So it can still work out cheaper than hosting on site. However the real value is a good hosting company take away the requirement for an on site engineer and may have 10-20 highly skilled exchange engineers that have a huge amount of experience between them to set up and mange the service. So new features are quickly implemented and down time is kept to a minimum.

As for the outage issues, that's some thing each company has to weigh up. But as a medium size company we have diverse links to separate points of presence to our ISP, and if you want to go further you can always go to two separate ISP's. In the 8 odd years I have worked in networking I have never had more than about 10 minutes of internet outage, which for non time critical office based work is not an issue.

I don't think you will see financial banks going cloud for there desktop or exchange any time soon. for both resilient and security reasons. But I think small business really can gain from it all.

phoeneous

Chivalry1 wrote: »

Understood Sizeon....just providing insight from my point of view as a consultant. With the cheap cost of ISP data couriers lines, which most companies already have, the risk of a network outage/slowness are mitigated. I know one client company that literally has 4 data lines to ensure business continuity. I know for some small companies in remote areas this won't be possible, but these days you can get a extra data line from a local ISP for less than $100 a month.

Yeah, until someone crashes a semi into the mpoe/demarc. Don't laugh, I've seen it happen.

Claymoore

dbrink wrote: »

You can easily build an exchange deployment for $400k that will last you a good 3-5 years.

20k users each with a 50GB primary mailbox is 1 Million GB or a Petabyte of storage. Each redundant copy in the DAG is another petabyte. That's a 1 PB you have to back up, so you need capacity and licensing for the backups. And that's just rough storage for primary mailboxes, we won't consider archiving or database overhead. Exchange has a limit of 2 TB per DB and a limit of 100 DB per server, which limits the DAG to 100 DB of 2 TB each for 200 TB. Now that's 5 separate DAGs you have to manage for your 1 PB of primary mailbox storage. If each is 3 servers - 2 in a primary DC and 1 in a secondary DC - that's 15 mailbox servers that require licensing and annual maintenance. Those servers have to rack somewhere, so that's monthly power/cooling/datacenter space fees. Plus you still have to buy Exchange CALs and probably Outlook for 20k users. That's likely part of your EA, but still not free.

Still think you can get all that for $400k?

dbrink

Claymoore wrote: »

20k users each with a 50GB primary mailbox is 1 Million GB or a Petabyte of storage. Each redundant copy in the DAG is another petabyte. That's a 1 PB you have to back up, so you need capacity and licensing for the backups. And that's just rough storage for primary mailboxes, we won't consider archiving or database overhead. Exchange has a limit of 2 TB per DB and a limit of 100 DB per server, which limits the DAG to 100 DB of 2 TB each for 200 TB. Now that's 5 separate DAGs you have to manage for your 1 PB of primary mailbox storage. If each is 3 servers - 2 in a primary DC and 1 in a secondary DC - that's 15 mailbox servers that require licensing and annual maintenance. Those servers have to rack somewhere, so that's monthly power/cooling/datacenter space fees. Plus you still have to buy Exchange CALs and probably Outlook for 20k users. That's likely part of your EA, but still not free.

Still think you can get all that for $400k?

You're right, I didn't factor in everything. We built ours for $400k for the Exchange hardware and storage. We have 2 copies of each database, one in each datacenter. We don't have 50GB mailboxes so our storage is nowhere near that. I didn't factor in the DPM storage we use to backup exchange so yeah that will be a bit more. So yeah it did cost more than $400k but is still cheaper than going to a hosted solution. We run about 30,000 mailboxes on 8 mailbox blades using direct attached storage and everything has been great for 2 years now.

DevilWAH

phoeneous wrote: »

Yeah, until someone crashes a semi into the mpoe/demarc. Don't laugh, I've seen it happen.

Or like us you have 1/2 Km between entry point and redundant comms rooms feeding it to site via multiply paths. the two connections take different paths out of site and to two separate exchanges.

Most ISP will feed your second link in via a different route if you request it, and as the whole point of getting a second link is for redundence then its some thing you should be factoring in. Even if you only have a single building you should bring the service in via separate routes and have it terminated in to a different location before being fed back. No point in getting a second line if it runs the same route with the same points of failure. most of the outages I have seen to networks are caused by physical damage, work men and Rats mostly. And compared to ongoing rental costs, asking for an alternative route in most cases are not that much.

it_consultant

I don't see on prem exchange going away but I think we will see a ton of hybrid deployments for all sorts of stuff including Exchange.

ptilsen

DevilWAH wrote: »

Most ISP will feed your second link in via a different route if you request it, and as the whole point of getting a second link is for redundence then its some thing you should be factoring in. Even if you only have a single building you should bring the service in via separate routes and have it terminated in to a different location before being fed back.

Not good enough, really. Unless you have multiple lines with carrier diversity, the chance of an outage is still pretty high. I mean, it depends on the industry and how much you can justify allocating to fault tolerance. For many organizations, the cost of running on-prem solutions is easily justified by the risk mitigated. In the US in particular, that cost is often lower than using cloud with any real fault tolerance on the WAN circuits. Our ISP markets are still pretty poor.

Of course, on the topic of Exchange, for most organizations keeping Exchange up when the WAN is down is probably not especially useful, email being an Internet-facing service. I'm just saying having physical line redundancy for one circuit is not necessarily going to be sufficient fault tolerance to justify putting a particular service elsewhere.

I'm with it_consultant on this one. On-prem is not going to disappear, but there will be relatively few pure on-prem solutions, and in many cases even those will only be done out of ignorance or irrationality, not because they're the most cost-effective solution. In the case of Exchange, specifically, Microsoft has really made it so you have to either have to have more outage risk or be a very, very large organization that can justify the economies of scale for a redundant, well-designed fully-on-prem Exchange solution (for reasons Claymoore explained).

DevilWAH

Umm I did mention in my original post the step up from multi links from one ISP, is diverse links to multiply ISP/Carriers. it is of course all a trade of between money and the amount of redundence you can get you can get. For me to have two completely separate diverse link from two ISP's, running to separate POP's, the insulation cost is about 50% incress of having them from the same ISP and same POP. So for medium enterprise companies the cost of getting full redundant links is not huge.

I would also say that having exchange up when the WAN is down is actually very important. 70%+ of our email traffic is internal emails. losing the external links currently we only would lose external email. move to cloud and we would lose both internal and external mail. Moving all services like VDi and Exchange to the cloud does mean that now if the exteranl links go down you can lose all ability to supply services to users.

Having said that I see hybrid as the way a lot of people will go.

GAngel

Did you factor in salaries (pensions,training, healthcare) much less all the other misc stuff that's add on value. I wouldn't be surprised if you were actually paying double all in. With a large setup like that you'd also be looking at 15-25% reductions from most players trying to get your business. It's never black and white in this industry.

No More SBS with exchange for essentials 2012 says they're trying to move the majority away from the physical infrastructure.

Chivalry1

phoeneous wrote: »

Yeah, until someone crashes a semi into the mpoe/demarc. Don't laugh, I've seen it happen.

As my Applied Statistics and Risk Management professor told me way back in college "Never deal in possibility.....always always in probability". Anything is possible, the question is whether it's probable. As a person who does regular risk analysis, this has helped me greatly. If your company has had a high rate of semi trucks running into the demarc point or the demarc stands the chance of being hit by vehicles...its time to re-evaluate the location of the demarc.

phoeneous

Chivalry1 wrote: »

As my Applied Statistics and Risk Management professor told me way back in college "Never deal in possibility.....always always in probability". Anything is possible, the question is whether it's probable. As a person who does regular risk analysis, this has helped me greatly. If your company has had a high rate of semi trucks running into the demarc point or the demarc stands the chance of being hit by vehicles...its time to re-evaluate the location of the demarc.

Yeah and with great risk analysis should also come great attention to detail; I never said it was where I worked.

RomBUS

My company just recently moved our Exchange/Outlook and CRM solutions to Microsoft's cloud services in the past. One frustrating thing I must say are the platform updates that take place during inopportune times (because it seems are off-site in a different time zone) and there's nothing we can do about it. Other than that it's been good in terms of speed and service

rsutton

sizeon wrote: »

Do you guys think on premise exchange is going to die out? With hosted exchange and office 365, i don't see why someone would want an inhouse exchange.

The reasons someone might never go with a hosted solution are:
-Cost
-Trust (e.g. trusting a 3rd party with your data)
-Ignorance / lack of understanding
-Control - some hosted providers do not give you the level of Control you may need. As an example, Intermedia only allows organizations to have one Transport rule per account. This is an example; every hosted provider I have worked with, including Office 365, does not give you the same amount of control as you would have with an on-premise solution

it_consultant

Ultimately it is up to the organization; if they are cloud amiable then Office 365 is a fantastic product, much better, IMHO than google apps. Realistically, even moderately sized Exchange deployments are not very expensive anymore. Exchange has gotten to the point where it is practically plug and play, I see very little need for dedicated Exchange engineers even for environments that have fairly large EX infrastructures.

Where we are really seeing a bit of a paradigm shift is in on-prem SQL. If your line of business application is buttoned down by SQL and you aren't considering Azure or AWS - you are wrong. If you want redundancy and load balancing and you aren't considering those two services you are also wrong.