For each user, Lightstreamer can allocate a maximum bandwidth that is dedicated to their streaming channel. For example, if it is required that a certain user cannot exceed a bandwidth of 10 kbps, Lightstreamer will filter the data in such a way as to ensure that the streaming connection of that user always remains below 10 kbps.

Lightstreamer can allocate a maximum update frequency for each user/item combination too. For example, it is possible to configure the profile of a certain user so that they do not receive more than 2 updates per second for a certain subscribed item (piece of data).

Filtering the data makes sense not only to limit the bandwidth occupied by the streaming connection, but also to improve the quality of data usage on the part of the end user. The "human interface" can absorb data by a combination of the human eye and brain, which are incapable of perceiving data changes of more than a few updates per second. Sending 30 updates a second would be a waste of bandwidth and computing resources, as they could never be taken in by a human user. However, Lightstreamer can also feed automated (as opposed to human) processes. In this case, it is possible to bypass the filtering mechanisms.

Lightstreamer provides several different subscription modes, which impact on data filtering. They are called MERGE, DISTINCT, COMMAND, and RAW and are flexible enough to cover all the types of data feed. The choice of the proper subscription mode depends on the nature of the data. See the White Paper to know more about the subscription modes.

The advantages of bandwidth control are obvious, for both the server and the client. For the server, bandwidth control makes it possible to size the Internet connection required simply and accurately. For the client, bandwidth control makes it possible to avoid saturating the bandwidth of the user's connection.

Lightstreamer automatically spots situations of congestion on the Internet, heuristically slowing down or suspending the dispatch of the data until the connection is again fully serviceable. This means that if at any time the actual available bandwidth is less than the allocated and required bandwidth, Lightstreamer uses its own filtering mechanisms to throttle the data flow. The advantage is that when the channel becomes available again, the user does not receive a burst of obsolete updates, but immediately starts to see the new data (in other words, data aging is avoided).

Adaptive streaming is particularly useful for connections that are notoriously unstable and variable, such as the mobile ones. Let's assume that the user starts a push session on their smartphone (using one of the many mobile clients made for Lightstreamer). If at a certain point the signal deteriorates and the effective bandwidth narrows, Lightstreamer Server dynamically adapts to the new network conditions, reducing the frequency of the updates, but still sending fresh real-time data.

Lightstreamer has direct control over the composition of each TCP packet. Instead of delegating the aggregation of data into packets to the operating system, Lightstreamer Server can impose its own batching policies, based on a trade-off between data latency and the reduction of the number of TCP packets.