docs/design: tiny update (#24166)

docs/design/2018-10-08-online-DDL.md

@@ -17,7 +17,7 @@ To simplify the design, the entire system allows only one node to make schema ch
 * **owner**: You can take it as a role. The information (including information of the node currently elected for this role) is stored in the Placement Driver (PD). Only one node in the entire system can be elected as the owner role. Each node can be elected to be this role, and workers will have the right to process jobs only when they are elected as the owner.
 * **workers**: Each node has 2 workers to process jobs. They will take the job from the corresponding job queue and process it. Workers will periodically check whether there are tasks that need to be processed. If the server receives the request while the owner role is on the same server, the start job module also tells the corresponding worker to process the job.
 * **meta data**: To simplify the design, we bring in system databases and system tables to record some metadatas in the process of asynchronous schema changes.
-* **state**: According to the change process of F1’s asynchronous schema, some states are introduced in this process, which are bound to column, index, table, and database. These states includes none, delete only, write only, write reorganization, and public. The former order is generally the order in which the operations are created. The state of the delete operation is the reverse of the creating order. Change the write reorganization to delete reorganization. Although they are all reorganization states, the visible level is different. So they are divided into two status flags.
+* **state**: According to the change process of asynchronous schema, some states are introduced in this process, which are bound to column, index, table, and database. These states includes none, delete only, write only, write reorganization, and public. The former order is generally the order in which the operations are created. The state of the delete operation is the reverse of the creating order. Change the write reorganization to delete reorganization. Although they are all reorganization states, the visible level is different. So they are divided into two status flags.
 * **lease**: At the same time, there are at most two different versions of the schema on the same table for all nodes of the system. That is, there are at most two different states. As a result, every normal node in a lease will automatically load schema information. If this node cannot be loaded normally during the lease, SQL cannot be processed normally.
 ## Rationale
 To dive deep into DDL processing, you need to understand the overall architecture of the DDL module. This module is operated on the TiDB Server, but it also involves the use of two other components of the TiDB cluster (PD Server and TiKV Server).