Issue 2 - Email Policies in the Age of DAOS

The revolutionary DAOS (Lotus Domino Attachment and Object Service) from IBM dramatically reduces the amount of server storage consumed by email by de-duplicating attachments stored on a single server.

Preliminary evidence shows a 40% reduction in storage is typical, with some people in the blogosphere claiming reductions of 60% or more. DAOS is a remarkable achievement by IBM.

This kind of radical performance improvement requires every company to review the impact of email policies..  For example, 10MB message size limits no longer yield the savings that they did before DAOS. Yet, the downsides of the policies, such as blocking critical email containing a large presentation, detailed photograph, or multimedia file to an important customer or vendor, remain.  Do the benefits of message size limits still outweigh the costs?

In addition, where are the opportunities to seize the next round of improvements? It is important to look at the fastest growing cost of email – the increasing volume of email messages. Each message, regardless of size, consumes overhead and wastes manpower. Personnel costs will become the most expensive part of an email system, now that storage costs are declining rapidly.

Basex, a frequently quoted New York-based research firm says that 25% of a knowledge worker’s day is consumed by too many messages and too many interruptions.

Network infrastructure can also be hit with outages and slow-downs due to the volume of email messages, regardless of how much storage they will eventually consume. The result is that service level commitments can be hit by areas not covered by DAOS or message size limits.

Where do these wasteful messages come from?

According to research by Permessa, based on 15 years of experience with leading global enterprises, 80% of email impact comes from just 4% of messages which, in turn, typically come from just 1% of the user community. Here are samples of email problems that could not be solved by DAOS or message size limits:

• The email servers at Total Oil Trading SA (TOTSA) stalled for minutes at a time and impacted service level commitments. It turned out that a few traders periodically sent broadcast emails with large attachments.
• A subscriber to the U.S. Department of Homeland Security’s “Open Source Intelligence Report” wanted to update his email address. Instead of sending his email to the administrator by hitting Reply, he accidentally hit Reply-to-All. His message was sent to the entire DHS distribution list. In the hours that followed, many subscribers responded, also using Reply-to-All. By the end of the day 7,500 DHS employees generated more than 2.2-million email responses that slowed email processing and filled mailboxes.
• Linde Gas, a European-based company with more than 16,000 employees, found its email volume increasing too quickly. The company determined that many workers sent 600-700MB of email to themselves each week for storage.

A detailed analysis of message traffic at large enterprises confirms that it is the volume of messages that has the biggest impact. The chart below, created by Permessa from real customer data, shows the actual message size distribution at enterprise installations and the resulting volume impact on the network.

As expected, the highest volume of messages consist of ordinary messages that are less than 100KB in size, as shown by the tallest blue bar. But, the total impact on the network is small, as shown on the adjacent gold bar.

Message Size Distribution and Network / Storage Impact

Extremely large messages, those bigger than 10MB, also have a relatively small impact on the network these days, as shown on the two small gold bars on the right side of the chart. This is because such messages are usually sent to small numbers of recipients and there are fewer of these messages overall. With DAOS taking care of the cost of duplicate storage, the impact of very large messages is even less.

The surprise to most organizations is the impact of messages between 1-5MB. These are often sent repeatedly to many recipients. As a result, these messages are the real burden on the infrastructure. (75% of the total data volume is caused by messages <5MB.) On the other hand, extremely large messages have a relatively small impact because they’re usually sent to small numbers of recipients and there are fewer of these messages overall.

Let’s look at policies that matter in the Age of DAOS.

Message Impact Limits

We already learned that the common 10MB message size limits, which were once critical in the days of dial-up connections and expensive storage with duplicate attachments, no longer yield significant savings. These size limits can also disrupt business activity. Presentations, detailed images, and multimedia files can easily exceed 10MB. Blocking or delaying these messages to a customer or vendor can delay or prevent deals from closing, without improving storage costs or network performance.

However, messages of 1-5MB that are sent to many people can actually have a significant impact on both the infrastructure and can interrupt users.

Permessa Email Control! measures the total impact of each message as it is sent. The message impact calculation is based on the number of recipients, the size of the message sent to each recipient, network topography and other factors. Message impact limits can be created and optimized for each enterprise.

Mailing Group Abuse

Abuse can occur when unauthorized workers use mailing groups to send inappropriate messages to large numbers of people. Such messages can easily reduce employee productivity on a massive scale. In the wrong hands, mailing groups and distribution lists can turn into a conduit for airing personal grievances.

Permessa Email Control! can limit the use of mailing groups to authorized personnel. Each group can be limited by job level, department, job function, or named senders.

Another issue with mailing groups is that once they are created, they rarely go away. Typically, corporate managers and executives make a request to IT in order to set up a new group. However, expiration dates are rarely set. As a result, groups accumulate over time, which makes it next to impossible for IT to manage or keep them current.

Permessa Email Control! solves this problem by identifying mailing groups that have not been used for an extensive period of time.

Tame Reply-to-All

The “Reply-to-All” button remains one of the biggest causes of internal email spam. A single message can be sent to 100 people. When some of the recipients reply to all in return, the one message can turn into dozens of disruptions with potentially useless content.

The problem can be solved easily with three simple techniques.

1. By limiting the use of mailing groups to authorized senders as mentioned above, the system can prevent replies sent by unauthorized users to an entire mailing group.
2. Create an enforceable email policy using Permessa Email Control! that limits the number of recipients in an email message. Restrictions may be set company-wide, but it may make more sense to limit users by their department or job level. For example, corporate officers and security personnel may deserve unlimited access.
3. Add a simple prompt to confirm before sending an enterprise-wide email. (Are you sure?) Selected email accounts, such as list servers and those used to send corporate newsletters may be exempted.

Conclusion

In the Age of DAOS, with reduced storage costs, it is important to look at other ways to reduce costs in email operation. Three simple steps can augment DAOS and will add to improving service and reducing costs:

1. Replace outdated message size limits with message impact limits.
2. Restrict the use of message groups (distribution lists) to authorized personnel.
3. Tame the Reply-to-All problem. Permessa Email Control! can easily help you to define and implement these programs to reduce cost and improve productivity.

Permessa Email Control! with Permessa Email Policy Enforcer automatically makes best practice recommendations based on email use and can help identify reply-to-all abuse patterns.  An optimal limit on the number of recipients can be placed on all messages or just those from certain senders.