[Okta Webinar] Learn how to a build a cloud-first strategyRegister Now

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 264
  • Last Modified:

Network Infrastructure - Windows 2003 Design

Our company is trying to move all our servers to a new collocation, the problem that I have run into is that I would really like to have 2 environments running at the same time so that we don't have to shut down all servers move them to Dallas, TX from Northern California and then hope everything goes well when we start things up there. We have a dedicated 10Mbps line between our office and the Collocation(Dallas, TX). We are going to have 1 web server, 3 Application Servers and 1 Database server. All the application servers are tied to eachother and the web server like a spider web unfortunately. We have all redundant servers that are available to us. Should I setup clustering and then run them simultaneously? I need some guidance if anyone has done this before. Thanks a million and I will reward the answer as many points as I can.
0
gbauer17
Asked:
gbauer17
  • 9
  • 8
1 Solution
 
redcelltechCommented:
Several questions

1. Do these machines host AD?

2. What is the age of your DR platforms in relation to your production equipment?

2. What is the time frame for migration?

With the answers to these questions we can begin to outline your migration plan.
0
 
Malli BoppeCommented:
Clustering is one option.But I wouldn't use MS clustering for globally spread servers.Heard that it won't work well on MS clustering better to use 3rd party clustering tools.

Even better option would be to use VMware clustering.I don't have much experience on Vmware just started working on it.I have heard that its best solution for DR or moving servers from one datacentreto a another.
0
 
gbauer17Author Commented:
redcelltech:

1. Yes we will have a dedicated Active Directory

2. I am not sure what "DR" platform is.

3. We want to migrate within 45 days
0
Prepare for your VMware VCP6-DCV exam.

Josh Coen and Jason Langer have prepared the latest edition of VCP study guide. Both authors have been working in the IT field for more than a decade, and both hold VMware certifications. This 163-page guide covers all 10 of the exam blueprint sections.

 
redcelltechCommented:
DR (Disaster recovery platforms) you spare units.

Microsoft clustering - to much to migrate and test within 45 days. This also does not meet your want of 2 running environments.

I have extensive experience with VMWare, a great product but if YOU do not have experience with it, I think the learning curve is to high for a 45 days migration timeframe. Again does not meet your want.

I think the simplest and probably the easiest to pull of would be as follows.

Assuming you have backup/recovery software that has and bare metal recovery tool. I would take the following steps.

Setup a seperate test network. I would backup the production servers and restore those images to the redundant hardware platforms. Over the next serveral weeks I would perform this several times to become proficient with the process. Also good experience if you have to do this for real.

This will give you the abilitiy to test the solution prior to shipping it and meet your want of two production networks. You could place domain workstation or clones of those machines into that environment and test

You will need a period of outage for this scenario. My thought is Friday close of business to Monday start of business (assuming this is possible). Snap a final image of the production server Friday after hours. Complete by Saturday morning. Put the production server on a major airline and ship them to your destination. While those machines are in transit, restore the final snapped images to your redundant platforms. This will be your fail over environment. Should something go wrong patching the redundant server environment into your production network and you are off and running.

This is just a rough outline obviously more thought needs to be given to the solution prior to action.

If for some reason your are not able to shut the production network down we can explore the possibility of a product like EMC's Replistor.  
0
 
gbauer17Author Commented:
redcelltech:

That is what I was originally going to do was to create an image of the servers Friday night and then put them on an airline for overnight and once they get there reimage and you are up and running. The problem I run into that the DNS will have to be changed with our ISP Name Servers, if they are changed to go to Dallas, TX and then have the Dallas site go down and then you would have to call and try and get those name servers changed back to point to our main office in Northern California. Why is it not possible to create clusters gerographically?
0
 
redcelltechCommented:
I my scenario, you have built the TX network prior to shipping it. Not imaging servers after delivery. If the set the TTL on your DNS records to one hour, Sunday you will be testing to insure everything is working. If for some reason they are not. You can have the records moved back by Monday.

As for clustering over WAN links. How much experience do you have with clustering in general. Clustering can be challenging when not up against a time frame. Based on your 45 day window, my thoughts are to much with to little time. I think the challenges of DNS records are easier to overcome than clustering 5 server pairs.

What version of server are you running?

What database server are you running?

Is the software the is running the application layer able to be clustered? Not all are, I have seen nasty issues with table locking and connection pooling in poorly written software.
0
 
gbauer17Author Commented:
We are running Windows 2003 Enterprise Server

We have SQL 2005.

We are using .Net code and know that the code is messy so that is one thing I am worried about is clustering actually causing more harm than good.
0
 
redcelltechCommented:
I think we have our answer regarding clustering then. It will take to much time to build the cluster pairs, only to find the .net code does not work well.

I still think the backup method is best for many reasons.

Allows for a testing prior to deployment.

Provides a easy go back direction should things go wrong. I have been thinking about the dns problem and I may have a solution. Follow me on this.

So if you were to take www (just and example) and point it to a host that was not part of the migration. This host would serve to do nothing but a browser redirect to www1 (the exsiting production site) or www2 (the new site in TX).  That way if things did not go well you could change on line of code on this redirect page to point to the old production network. We you know the solution is up and running you just migrate away from this scenario. This allows the ohhhhh no we have to go back scenario.

0
 
gbauer17Author Commented:
redcelltech:

So you think that the best solution is to create an image of the the production servers right now and then put that image on the servers that will be gong to TX. Ship them overnight to the location and then start them up. What if any data has changed in that 14hrs? Or should we setup that redirect page to say that the site is down for maintenance so that will prohibit any changes from being made? Then if the TX site is up and running then you are golden.
0
 
redcelltechCommented:
Yes, the redirect page is definitely part of this equation. As for the page message, I think it is a great opportunity to tell your customers you are moving your severs to a better COLO facility that they will benefit from and give the numbers to call for sales, if applicable.
0
 
gbauer17Author Commented:
How would you suggest to create the image? Ghost? Or is there something else out there that will do a better job? Also what if the equipment is a little different?
0
 
redcelltechCommented:
I don't know what you are using for system backups now, but you could start there. Most major backup software vendors have a solution for doing this. Some are easier to use than others.

Here is a product I really like, and will allow you to restore an image to different hardware. It is also inexpensive relative to some other products.

http://www.acronis.com/enterprise/products/ATIESWin/

I can help you with making the some of these choices if needed.
0
 
gbauer17Author Commented:
We currently run Backup Exec but I have not known it to provide a way to backup a image of a server. I have heard of Acronis before so maybe I can use this to create a image of each server before shut down and taken to collocation facility. I appreciate your offer to help with some choices and might have to take you up on this offer. I have never done a project like this before but you always have to start somewhere.
0
 
redcelltechCommented:
I run an outsourced IT company providing support to companies across the country. You can contact me at tll@redcelltechnologies.com, and I will send you my cell phone number. I think based on your deadline, a conference call will speed this process along.
0
 
redcelltechCommented:
How are you coming along with the project?
0
 
gbauer17Author Commented:
Unfortunately we are awaiting some approvals from the higher ups. I am a consultant in a small city and I am waiting for the IT Manager to meet with me on Monday to go over some things. I am pretty confident that we are going to go with the option of sending over our servers and cut in 1 weekend. Then once this is in place then we will go ahead with clusters. How much do you charge?
0
 
gbauer17Author Commented:
Not sure if you know or not but your website is not working.
0
 
redcelltechCommented:
We are in process of updating our website. We are tying Kayako support suite into it and making many other changes.
0

Featured Post

Industry Leaders: We Want Your Opinion!

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

  • 9
  • 8
Tackle projects and never again get stuck behind a technical roadblock.
Join Now