LuxSci provisions customers on dedicated servers with plenty of "horsepower" to accommodate each customer's stated sending throughput requirements. For example, effectively sending 100,000 messages in a few minutes instead of over the course of a day requires the use of a cluster of servers instead of a single dedicated server. LuxSci's sales engineers will design a custom solution that meets your throughput and high availability needs.
When delivery speed (i.e., messages/minute or messages/hour) is a factor, a more powerful server will receive, process, and deliver deliver your messages faster and will allow you to use more sustained concurrent connections. The following table indicates how fast servers of different sizes can process and send SecureLine-encrypted email messages. I.e., this is the maximum throughput under ideal conditions. LuxSci likes to provision customers so that the maximum throughput is at least twice the customer's stated capacity requirement.
Dedicated Server Size
|1 CPU||2 CPU||4 CPU||8 CPU|
The maximum sustained rate at which you can deliver email to the server, have it processed, and have it delivered to the recipient's mail servers depends on many factors. These include:
The speed by which you can deliver message to your server:
The speed by which the messages can be processed (i.e., encrypted, tracked, etc.):
The speed by which the messages can be delivered to recipients
These factors determine how fast you can have a quantity of messages delivered to their recipients. Some of these factors are under your control and some of them are not.
The "maximum sending rate" (msgs/hr) is the upper bound rate that you can achieve through the service. It assumes optimal sending conditions for high volume encrypted email: very small (5 KB) messages, all to the same domain, sent over a fast network using SMTP Pipelining, and encrypted using TLS. Under these conditions, there is no block listing, gray listing, or DNS slowness, there are no problems with some recipient email servers, and we choose an optimal sustained number of concurrent sending connections to the server.
|Maximum msgs/day||This is an estimate of the maximum number of messages that your server could process and deliver over 24 hours of sustained sending at approximately the maximum hourly rate.|
For some customers, the speed of sending is very important. There are two significant modes of sending:
Making one or more connections to your server and then sending at a consistent rate for a long period of time. E.g. 20,000 messages/hour for minutes or hours. Sustained rates are relevant when sending an email marketing blast to a large subscriber list or when offloading a large number of messages from an external application through your sending server.
A "burst" refers to sending a bunch of messages in a very short time; e.g. 25 or 50 messages within a second or two. Burst differs from "sustained" in that "bursts" are short duration events and when you are not "bursting" you are sending relatively few messages here and there ... infrequent sending. Bursting is relevant for transactional email messages when you are sending them "as needed" but require the ability to send many at once should that need arise, as it will.
Sending speed is important for customers who need to send large mailings at a sustained rate. Understanding sending speed helps you determine how long it will take to send that blast to 100,000 or 1 million subscribers.
There is a theoretical maximum number of messages/hour that any server can manage. How fast you can actually send is affected by a number of factors.
|1. Server Power||
More powerful servers can handle more concurrent connections, process more messages at once, deliver more messages at once, and thus give you higher sending rates. The number of CPU cores is the primary factor governing performance; however, memory and disk speed are also important.
|2. Email Sending Program||
Different email marketing/sending programs have different capacities for how efficiently they can send high volumes of email. Simply switching from one program to another can increase your throughput significantly.
|3. SMTP Pipelining||
Use an email marketing program that supports SMTP Pipelining. With this feature, the program opens a single authenticated connection to the server and sends multiple messages in succession over that connection. This is significantly faster than opening a new connection and authenticating again for every single message to be sent.
|4. Concurrent Connections||
If your sending program can open multiple connections to our SMTP servers at the same time and send multiple messages at once, then you can delivery messages to our servers more quickly. Depending on how powerful your server is and how slow your network connections are, you can use more and more concurrent connections to achieve higher and higher sending rates. Combined with SMTP Pipelining, an appropriate number of concurrent connections goes a long way to maximizing your throughput potential.
Under most circumstances, sending programs can deliver messages to our servers much more quickly than the servers can process those messages (i.e., encrypt them, track them, log them, etc.) and then deliver them to the recipients' mail servers.
The recommended maximum number of concurrent connections to use for sustained email sending is given in the table, above. It is important to note that while exceeding that recommended value will get your email to your server faster, sustained sending at that rate will overtax the server and cause the overall rate of processing and delivery of your messages to your recipients to slow down. Stick to the recommended number of concurrent connections for sustained sending.
|5. Network Speed||
Your messages must be transmitted from your email sending program to our servers. The speed of your network, any packet loss, and the distance between your sending computer and our servers can significantly impact the maximum rate that you can send messages.
|6. Message Size||
Like network speed, the larger your messages are, the more time it will take to transmit them from your email sending program to our servers. For example, on a fast network, increasing message size from 5KB to 1MB decreased throughput by 50%. The actual impact of size will depend greatly on your network speed and other factors.
|7. Recipient Email Server and DNS Speed||
In order for our servers to deliver messages to your recipients, the servers have to (a) look up where to deliver the messages to (DNS), and (b) talk to those servers to deliver the messages to them. If the DNS lookup as slow or the recipients are in domains that do not exist, this step can take a while. If the recipient's email servers are overloaded, on a slow network, or are having other issues, then the delivery phase can be slow.
Delivering many messages to recipients with these issues can slow down your entire sending process... as other messages have to wait for the slower ones to be sent.
|8. Block listing and Greylisting||
If your email content or sending patterns have gotten your server block listed or greylisted, then delivering email messages to recipient servers that are now blocking you or deferring you can slow down your delivery rates.
This can also cause a backlog of messages that are waiting to be accepted by these recipient servers. The need to Retry the backlog slows the process of delivering all other messages (though it does not block their delivery).
|9. Recipients per message||
Most email marketing and transactional email messages are sent individually to each recipient. This allows easier tracking of opens, views, etc. However, if you send each message to many recipients (e.g., 100s of recipients), then your sending program has significantly fewer messages to deliver to our servers. The delivery of your messages to our servers can happen 10s or 100s of times faster than if you sent in the usual one recipient/message pattern.
Additionally, if each message contains only recipients in a specific domain, then actual delivery to these recipients can be expedited by this efficiency. E.g. If you are sending to 500 AOL users, instead of sending 500 individual messages, send one message with all 500 AOL users listed as "BCC" recipients. Delivery can be almost 500 times faster.
|10. SecureLine Email Encryption||
In the example sending rates, we assumed that all messages would be sent using SecureLine TLS. If PGP, S/MIME, or Escrow are used, that can also impact email processing speed.
|11. Outbound Email Processing||
If you are using other email processing features that come with High Volume Secure Sending (e.g. content scanning, recipient restrictions, size restrictions,etc.), this processing may affect your maximum sending rate.
LuxSci has features that can be used to rate limit the processing of outbound email. This is usually used when "warming up" IP addresses to assist senders in sending slowly and establishing their IP address reputation.
This rate limiting is on, your maximum sending rate will be capped by this process.