# Low-latency with data-availability for Tezos

The ambitious goal of achieving 1 million transactions per second (TPS) for Tezos through smart rollups requires an effective data-availability solution.

Presently, two prominent solutions exist:

- Data-availability Committee
- Data-availability Layer

The primary inquiry we address in this document is: how can we achieve low latency (under a second) while employing a data-availability solution?

The prevalent solution currently is to rely on a sequencer. A sequencer is an entity that determines the order in which operations must be executed.

In the context of smart rollups, execution can occur at a different time from when the operation is processed from the inbox. Indeed, while operations must be read sequentially from the inbox in the order dictated by Layer-1 (L1), it is up to the kernel to decide in which order these operations can be actually executed.

Low latency can be achieved if the sequencer is the sole entity processing a rollup's transactions. The sequencer can proactively process all known operations, generate the result, and deliver it to the user. If additional operations exist in the inbox, the sequencer simply needs to read and execute them at a later time. In essence, low-latency relies on the fact that importing two operations into a rollup's local storage can occur in any order, as long as the execution sequence of these two operations is fixed.

:::warning
Here, we are making a simplification, as we're also assuming that the internal messages automatically added to the inbox will not influence the outcome of an operation.
:::

When a data-availability solution is introduced, it only alters the source from which these operations are imported. The sequencer still retains the capacity to decide the execution order of operations, even though data-availability comes with inherent latency (like the DAL). It's the sequencer's responsibility to ensure that this data is indeed published and flagged as available by the L1.

Another key difference between the DAC and the DAL is that, most of the time, the DAL doesn't need to use the shared inbox, as the kernel knows which slot it should retrieve data from on the DAL.

Therefore, the question of low-latency appears to be less related to the data-availability solution (DAC/DAL) and more linked to the presence of a single sequencer for the rollup, which decides the sequence of operations.