Insert Entities

This topic describes how to insert data in Milvus via client.

You can also migrate data to Milvus with MilvusDM, an open-source tool designed specifically for importing and exporting data with Milvus.

Milvus 2.1 supports the VARCHAR data type on scalar fields. When building indexes for VARCHAR-type scalar fields, the default index type is dictionary tree.

The following example inserts 2,000 rows of randomly generated data as the example data (Milvus CLI example uses a pre-built, remote CSV file containing similar data). Real applications will likely use much higher dimensional vectors than the example. You can prepare your own data to replace the example.

Prepare data

First, prepare the data to insert. Data type of the data to insert must match the schema of the collection, otherwise Milvus will raise exception.

Once you enable dynamic schema, you can append dynamic fields in the data. For details, refer to Dynamic Schema.

import random
data = [
  [i for i in range(2000)],
  [str(i) for i in range(2000)],
  [i for i in range(10000, 12000)],
  [[random.random() for _ in range(2)] for _ in range(2000)]
]

## Once your collection is enabled with dynamic schema,
## you can add non-existing field values.
data.append([str("dy"*i) for i in range(2000)])

const data = Array.from({ length: 2000 }, (v,k) => ({
  "book_id": k,
  "word_count": k+10000,
  "book_intro": Array.from({ length: 2 }, () => Math.random()),
}));

bookIDs := make([]int64, 0, 2000)
wordCounts := make([]int64, 0, 2000)
bookIntros := make([][]float32, 0, 2000)
for i := 0; i < 2000; i++ {
	bookIDs = append(bookIDs, int64(i))
	wordCounts = append(wordCounts, int64(i+10000))
	v := make([]float32, 0, 2)
	for j := 0; j < 2; j++ {
		v = append(v, rand.Float32())
	}
	bookIntros = append(bookIntros, v)
}
idColumn := entity.NewColumnInt64("book_id", bookIDs)
wordColumn := entity.NewColumnInt64("word_count", wordCounts)
introColumn := entity.NewColumnFloatVector("book_intro", 2, bookIntros)

Random ran = new Random();
List<Long> book_id_array = new ArrayList<>();
List<Long> word_count_array = new ArrayList<>();
List<List<Float>> book_intro_array = new ArrayList<>();
for (long i = 0L; i < 2000; ++i) {
	book_id_array.add(i);
	word_count_array.add(i + 10000);
	List<Float> vector = new ArrayList<>();
	for (int k = 0; k < 2; ++k) {
		vector.add(ran.nextFloat());
	}
	book_intro_array.add(vector);
}

var bookIds = new long[2000];
var wordCounts = new long[2000];
var bookIntros = new ReadOnlyMemory<float>[2000];

for (var i = 0; i < 2000; i++)
{
    bookIds[i] = i;
    wordCounts[i] = i + 10000;
    bookIntros[i] = new[] { Random.Shared.NextSingle(), Random.Shared.NextSingle() };
}

# See the following step.

Insert data to Milvus

Insert the data to the collection.

By specifying partition_name, you can optionally decide to which partition to insert the data.

from pymilvus import Collection
collection = Collection("book")      # Get an existing collection.
mr = collection.insert(data)

const mr = await milvusClient.insert({
  collection_name: "book",
  fields_data: data,
});

_, err = milvusClient.Insert(
	context.Background(), // ctx
	"book",               // CollectionName
	"",                   // partitionName
	idColumn,             // columnarData
	wordColumn,           // columnarData
	introColumn,          // columnarData
)
if err != nil {
	log.Fatal("failed to insert data:", err.Error())
}

List<InsertParam.Field> fields = new ArrayList<>();
fields.add(new InsertParam.Field("book_id", book_id_array));
fields.add(new InsertParam.Field("word_count", word_count_array));
fields.add(new InsertParam.Field("book_intro", book_intro_array));

InsertParam insertParam = InsertParam.newBuilder()
  .withCollectionName("book")
  .withPartitionName("novel")
  .withFields(fields)
  .build();
milvusClient.insert(insertParam);

await milvusClient.GetCollection("book").InsertAsync(new FieldData[]
{
    FieldData.Create("book_id", bookIds),
    FieldData.Create("word_count", wordCounts),
    FieldData.CreateFloatVector("book_intro", bookIntros)
});

# insert an entity to a collection
curl -X 'POST' \
  '${MILVUS_HOST}:${MILVUS_PORT}/v1/vector/insert'  \
  -H 'Authorization: Bearer ${TOKEN}' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
         "collectionName": "collection1",
         "data": {
             "vector": [0.1, 0.2, 0.3],
             "name": "tom",
             "email": "tom@zilliz.com",
             "date": "2023-04-13"
          }
     }'

# insert multiple entities
curl --request POST \
     --url '${MILVUS_HOST}:${MILVUS_PORT}/v1/vector/insert' \
     --header 'Authorization: Bearer <TOKEN>' \
     --header 'accept: application/json' \
     --header 'content-type: application/json' \
     -d '{
         "collectionName": "collection1",
         "data": [
             {
                "vector": [0.1, 0.2, 0.3],
                "name": "bob",
                "email": "bob@zilliz.com",
                "date": "2023-04-13"
             },{
                "vector": [0.1, 0.2, 0.3],
                "name": "ally",
                "email": "ally@zilliz.com",
                "date": "2023-04-11"
             }
         ]
     }'

{
    "code": 200,
    "data": {
        "insertCount": "integer"
    }
}

Flush the Data in Milvus

When data is inserted into Milvus, it is stored in segments. Segments have to reach a certain size before they can be sealed and indexed. Unsealed segments are searched using brute force. If you need to search the data immediately after insertion, you can call the flush() method once the data is inserted. This method seals any remaining segments and sends them for indexing. It is important to only call this method at the end of an insert session. Calling it too frequently will result in fragmented data that will require cleaning later on.

Limits

What's next

• Learn more basic operations of Milvus:
  • Build an index for vectors
  • Conduct a vector search
  • Conduct a hybrid search