Cloud vs sand: Google vs Microsoft
The numbers don't lie An independent study found on-site Microsoft apps - Office and Exchange - cost 20x in capital dollars and 5x-6x more than Google Apps on a 3 year Total Cost of Ownership (TCO) basis. How can Microsoft compete?
The debate Cloud, as in Google apps, and sand, as in locally hosted Microsoft apps, are battling for business mind share. "Cloud is cheaper" say proponents. "Traditional apps are more reliable" say skeptics.
The rub: both are right. The business problem is finding the most cost-effective path given your needs.
No Microsoft or Google money A Boston company TwinStrata did this study with no funding from Google or Microsoft. TwinStrata sells Clarity AP (Assessment & Planning), that quantifies infrastructure data loss and downtime risks.
Figuring out downtime People are lousy at estimating the risk of uncommon events. All disk drives fail, yet few back up. SATA RAID 5 is no longer safe, yet people buy it.
But when one solution costs 20x more, it makes sense to take a 2nd look. Clarity AP uses Bayesian analysis to model the reliability and availability of systems, software, networks and operations. You can compare the difference between RAID 5 and 6, or tape and disk backup, or a Tier 1 data center against Amazon's S3. And much more, such as recovery processes.
Bayesian methods determine the total variability of a group of many subsystems. Virtually forgotten 40 years ago the technique is now widely used. (Here's a short YouTube video on Bayesian theory and the software.)
Clarity AP figures out the expected availability of a complex system and, by plugging in costs for downtime and data loss, figure out the expected uptime of a configuration and what the downtime cost.
The study focuses on cost and data availability. It assumes that application integration, security, performance and compliance meets minimum business requirements.
It also assumes that you have about 20 employees and use good quality products, such as a NetApp external filer.
The method
Model assumptions
| Daily volume of new email | 100 MB |
| Local copy of email | None |
| Archiving | None |
| Cost per hour of downtime | $500 |
| Cost per GB of data loss | $5,000 |
| Network outage | Lose email access |
Office apps assumptions:
| Daily volume of new documents | 50 MB |
| Local copies of documents | None |
| Cost per hour of downtime | $250 |
| Cost per GB of data loss | $30,000 |
| Network outage | No application access |
The cost of downtime and data loss is critical for assessing cloud vs sand. The Clarity AP software makes it easy to perform sensitivity analysis.
The results The study looked at four different configurations: 1) Google apps with a single network connection; 2) Google apps with two network connections for added availability; 3) Microsoft Office and Exchange with internal disk storage; and, 4) Office and Exchange with an external storage array.
| Solution | Capital expense | Operating expense | 3 year TCO |
| $1.3k | $10.7k | $33.4k | |
| Google dual network | $1.3k | $17.4k | $53.5k |
| MS internal disk | $27.5k | $40.5k | $148.9k |
| MS external array | $69.1k | $46.5k | $208.8k |
What about risk? Most new small businesses would stop there: 1/20th the cost of an on-site system is too good to ignore. But what if you are thinking of migrating to Google or your existing system needs replacement?
Here's how the options cost out when adjusted for risk.
| Solution | Annual risk of downtime | Annual risk of data loss | Adjusted TCO |
| $19.1k | $1k | $93.5k | |
| Google dual network | $13.2k | $1k | $95.9k |
| MS internal disk | $10.2k | $8.6k | $205.3k |
| MS external array | $7.2k | $1.1k | $233.7k |
Adjusting for the cloud's greater downtime Google is less than half the cost of Microsoft. As Google's availability improves the cost advantage will only grow.
If Google ever figures out how to make their stuff usable by small business owners, Redmond will be a ghost town.
The Storage Bits take Google will, of course, blow their huge cost advantage over Microsoft. At heart they're geeks who don't understand small business. In 20 years they'll look like Sun does now.
Here are a few of Google's problems in the SMB arena.
- Privacy suckage. They ask for too much information and aren't explicit about how information will be used. For example, if you want to use Google calendar you have to upload all your contacts to Google. Why?
- Non-existent support. People are willing to pay for support. Google needs to figure it out. They could, for example, train and credential ISP, VARs and consultants to offer Google Apps support.
- User experience and design. Google's homepage, designed for minimum bandwidth when most people used modems, was an accidental triumph - but it's been downhill from there. Marissa Mayer, VP of User Ignorance, is a disaster. Marissa, a word of advice: you're young, rich and w-a-a-ay too comfortable at Google - go spread your wings at another company and find out if you're smart or just lucky. Hint: it's the latter.
The real message of this analysis is that the economics of the Internet have made it a competitive advantage to be small. Capital requirements are minimal - good thing today - and cloud infrastructure + cheap local computes and storage
Comments welcome, of course. I've been doing some work for TwinStrata - like the video mentioned above - but the study was their idea. Update:Here's a link to a 3MB zip of a pdf presentation on the study from StorageMojo.com. End update.