Opinion: Why We Replaced Elasticsearch 8.15 with Meilisearch 1.5 for 60% Faster Next.js 15 Search

// app/api/search/route.ts\n// Next.js 15 App Router API route for Meilisearch 1.5 product search\n// Dependencies: @meilisearch/instant-meilisearch@0.12.0, meilisearch@1.5.0\nimport { NextRequest, NextResponse } from 'next/server';\nimport { Meilisearch } from 'meilisearch';\n\n// Initialize Meilisearch client with error handling for missing env vars\nlet meiliClient: Meilisearch;\ntry {\n  if (!process.env.MEILI_HOST || !process.env.MEILI_MASTER_KEY) {\n    throw new Error('Missing required Meilisearch environment variables: MEILI_HOST, MEILI_MASTER_KEY');\n  }\n  meiliClient = new Meilisearch({\n    host: process.env.MEILI_HOST,\n    apiKey: process.env.MEILI_MASTER_KEY,\n  });\n} catch (initError) {\n  console.error('Failed to initialize Meilisearch client:', initError);\n  // Throw to prevent unhandled client usage\n  throw new Error(`Meilisearch client initialization failed: ${initError instanceof Error ? initError.message : 'Unknown error'}`);\n}\n\n// Define typed product document interface for type safety\ninterface ProductDocument {\n  id: string;\n  name: string;\n  description: string;\n  price: number;\n  category: string;\n  in_stock: boolean;\n  created_at: string;\n}\n\n// Index name for product search\nconst PRODUCT_INDEX = 'products_v1';\n\n/**\n * Handle GET requests to /api/search?q=&category=&page=\n * Returns paginated search results with filters, typed to ProductDocument\n */\nexport async function GET(request: NextRequest) {\n  try {\n    // Extract search parameters with defaults\n    const searchParams = request.nextUrl.searchParams;\n    const query = searchParams.get('q') || '';\n    const category = searchParams.get('category') || undefined;\n    const page = parseInt(searchParams.get('page') || '1', 10);\n    const limit = parseInt(searchParams.get('limit') || '20', 10);\n\n    // Validate pagination parameters\n    if (isNaN(page) || page < 1) {\n      return NextResponse.json(\n        { error: 'Invalid page parameter: must be a positive integer' },\n        { status: 400 }\n      );\n    }\n    if (isNaN(limit) || limit < 1 || limit > 100) {\n      return NextResponse.json(\n        { error: 'Invalid limit parameter: must be between 1 and 100' },\n        { status: 400 }\n      );\n    }\n\n    // Get product index, handle missing index error\n    let productIndex;\n    try {\n      productIndex = meiliClient.index(PRODUCT_INDEX);\n      // Verify index exists by fetching stats (throws if index missing)\n      await productIndex.getStats();\n    } catch (indexError) {\n      console.error(`Product index ${PRODUCT_INDEX} not found:`, indexError);\n      return NextResponse.json(\n        { error: 'Search index unavailable. Please try again later.' },\n        { status: 503 }\n      );\n    }\n\n    // Build search options with filters and pagination\n    const searchOptions = {\n      page,\n      limit,\n      filter: category ? `category = \"${category}\"` : undefined,\n      attributesToRetrieve: ['id', 'name', 'price', 'category', 'in_stock'],\n      attributesToHighlight: ['name', 'description'],\n    };\n\n    // Execute search with error handling\n    const searchResults = await productIndex.search(query, searchOptions);\n\n    // Transform results to client-safe format (omit internal fields)\n    const transformedResults = {\n      total: searchResults.estimatedTotalHits,\n      page: searchResults.page,\n      limit: searchResults.limit,\n      results: searchResults.hits.map((hit) => ({\n        id: hit.id,\n        name: hit.name,\n        price: hit.price,\n        category: hit.category,\n        in_stock: hit.in_stock,\n        highlighted_name: hit._highlightResult?.name?.value || hit.name,\n      })),\n    };\n\n    // Set cache headers for Next.js 15 ISR (cache for 60 seconds, revalidate in background)\n    return NextResponse.json(transformedResults, {\n      headers: {\n        'Cache-Control': 'public, s-maxage=60, stale-while-revalidate=120',\n      },\n    });\n  } catch (searchError) {\n    // Catch-all error handling for unexpected failures\n    console.error('Search API error:', searchError);\n    return NextResponse.json(\n      { error: 'Failed to execute search. Please try again.' },\n      { status: 500 }\n    );\n  }\n}\n

// app/api/search/es-legacy/route.ts\n// Legacy Elasticsearch 8.15 search API route (pre-migration)\n// Dependencies: @elastic/elasticsearch@8.15.0\nimport { NextRequest, NextResponse } from 'next/server';\nimport { Client as ElasticsearchClient } from '@elastic/elasticsearch';\n\n// Initialize Elasticsearch client with auth and error handling\nlet esClient: ElasticsearchClient;\ntry {\n  if (!process.env.ELASTICSEARCH_NODE || !process.env.ELASTICSEARCH_API_KEY) {\n    throw new Error('Missing required Elasticsearch environment variables: ELASTICSEARCH_NODE, ELASTICSEARCH_API_KEY');\n  }\n  esClient = new ElasticsearchClient({\n    node: process.env.ELASTICSEARCH_NODE,\n    auth: {\n      apiKey: process.env.ELASTICSEARCH_API_KEY,\n    },\n    // Required for Next.js edge compatibility (disable gzip for edge runtime)\n    maxResponseSize: 10000000,\n    requestTimeout: 5000,\n  });\n} catch (initError) {\n  console.error('Failed to initialize Elasticsearch client:', initError);\n  throw new Error(`Elasticsearch client initialization failed: ${initError instanceof Error ? initError.message : 'Unknown error'}`);\n}\n\n// Elasticsearch index name for products\nconst ES_PRODUCT_INDEX = 'products_v1';\n\n// Typed product source for Elasticsearch hits\ninterface ESProductSource {\n  id: string;\n  name: string;\n  description: string;\n  price: number;\n  category: string;\n  in_stock: boolean;\n  created_at: string;\n}\n\n/**\n * Handle GET requests to /api/search/es-legacy?q=&category=&page=\n * Equivalent functionality to Meilisearch route, with Elasticsearch-specific overhead\n */\nexport async function GET(request: NextRequest) {\n  try {\n    const searchParams = request.nextUrl.searchParams;\n    const query = searchParams.get('q') || '';\n    const category = searchParams.get('category') || undefined;\n    const page = parseInt(searchParams.get('page') || '1', 10);\n    const limit = parseInt(searchParams.get('limit') || '20', 10);\n\n    // Validate parameters (same as Meilisearch route)\n    if (isNaN(page) || page < 1) {\n      return NextResponse.json(\n        { error: 'Invalid page parameter: must be a positive integer' },\n        { status: 400 }\n      );\n    }\n    if (isNaN(limit) || limit < 1 || limit > 100) {\n      return NextResponse.json(\n        { error: 'Invalid limit parameter: must be between 1 and 100' },\n        { status: 400 }\n      );\n    }\n\n    // Calculate Elasticsearch from/size for pagination\n    const from = (page - 1) * limit;\n\n    // Build Elasticsearch search query with bool filter and multi_match\n    const esQuery = {\n      index: ES_PRODUCT_INDEX,\n      from,\n      size: limit,\n      query: {\n        bool: {\n          must: [\n            // Multi-match query across name and description if query is provided\n            ...(query ? [{\n              multi_match: {\n                query,\n                fields: ['name^3', 'description'], // Boost name field by 3x\n                fuzziness: 'AUTO',\n              },\n            }] : []),\n          ],\n          filter: [\n            // Category filter if provided\n            ...(category ? [{\n              term: { category },\n            }] : []),\n            // Only return in-stock products\n            {\n              term: { in_stock: true },\n            },\n          ],\n        },\n      },\n      // Return only required fields to reduce payload size\n      _source: ['id', 'name', 'price', 'category', 'in_stock'],\n      // Highlight matching fields\n      highlight: {\n        fields: {\n          name: {},\n          description: {},\n        },\n      },\n    };\n\n    // Execute Elasticsearch search with timeout and error handling\n    let esResponse;\n    try {\n      esResponse = await esClient.search(esQuery);\n    } catch (esError) {\n      console.error('Elasticsearch search failed:', esError);\n      // Handle index not found error specifically\n      if (esError instanceof Error && esError.message.includes('index_not_found_exception')) {\n        return NextResponse.json(\n          { error: 'Search index unavailable. Please try again later.' },\n          { status: 503 }\n        );\n      }\n      throw esError; // Re-throw for catch-all handler\n    }\n\n    // Transform Elasticsearch results to match Meilisearch response shape\n    const transformedResults = {\n      total: esResponse.hits.total?.value || 0,\n      page,\n      limit,\n      results: esResponse.hits.hits.map((hit) => ({\n        id: hit._source?.id || hit._id,\n        name: hit._source?.name || '',\n        price: hit._source?.price || 0,\n        category: hit._source?.category || '',\n        in_stock: hit._source?.in_stock || false,\n        highlighted_name: hit.highlight?.name?.[0] || hit._source?.name || '',\n      })),\n    };\n\n    // Cache headers (shorter cache because Elasticsearch updates are slower)\n    return NextResponse.json(transformedResults, {\n      headers: {\n        'Cache-Control': 'public, s-maxage=30, stale-while-revalidate=60',\n      },\n    });\n  } catch (searchError) {\n    console.error('Elasticsearch search API error:', searchError);\n    return NextResponse.json(\n      { error: 'Failed to execute search. Please try again.' },\n      { status: 500 }\n    );\n  }\n}\n

// app/api/cron/index-products/route.ts\n// Next.js 15 cron job to index products into Meilisearch 1.5\n// Runs every 15 minutes via Vercel Cron or EC2 cron\n// Dependencies: meilisearch@1.5.0, @vercel/postgres@0.8.0\nimport { NextRequest, NextResponse } from 'next/server';\nimport { Meilisearch } from 'meilisearch';\nimport { sql } from '@vercel/postgres';\n\n// Initialize Meilisearch client (reuse from search route, but re-init for isolation)\nconst meiliClient = new Meilisearch({\n  host: process.env.MEILI_HOST!,\n  apiKey: process.env.MEILI_MASTER_KEY!,\n});\n\nconst PRODUCT_INDEX = 'products_v1';\nconst BATCH_SIZE = 1000; // Index 1000 products per batch to avoid memory issues\n\n/**\n * Handle POST requests from cron job to reindex products\n * Protected by Vercel Cron secret to prevent unauthorized access\n */\nexport async function POST(request: NextRequest) {\n  try {\n    // Verify cron secret to prevent unauthorized indexing requests\n    const cronSecret = request.headers.get('x-vercel-cron-secret');\n    if (cronSecret !== process.env.CRON_SECRET) {\n      return NextResponse.json(\n        { error: 'Unauthorized: Invalid cron secret' },\n        { status: 401 }\n      );\n    }\n\n    // Get product index, create if it doesn't exist\n    let productIndex;\n    try {\n      productIndex = meiliClient.index(PRODUCT_INDEX);\n      await productIndex.getStats();\n    } catch (indexError) {\n      // Index doesn't exist, create it with settings\n      console.log(`Creating new Meilisearch index: ${PRODUCT_INDEX}`);\n      const createResponse = await meiliClient.createIndex(PRODUCT_INDEX, {\n        primaryKey: 'id',\n      });\n      // Wait for index creation to complete\n      await meiliClient.waitForTask(createResponse.taskUid);\n      productIndex = meiliClient.index(PRODUCT_INDEX);\n\n      // Configure index settings for optimal search performance\n      const settingsTask = await productIndex.updateSettings({\n        searchableAttributes: ['name', 'description', 'category'],\n        filterableAttributes: ['category', 'in_stock', 'price'],\n        sortableAttributes: ['price', 'created_at'],\n        rankingRules: [\n          'words',\n          'typo',\n          'proximity',\n          'attribute',\n          'sort',\n          'exactness',\n        ],\n      });\n      await meiliClient.waitForTask(settingsTask.taskUid);\n    }\n\n    // Fetch total product count to calculate batches\n    const countResult = await sql`SELECT COUNT(*) FROM products WHERE active = true`;\n    const totalProducts = parseInt(countResult.rows[0].count, 10);\n    const totalBatches = Math.ceil(totalProducts / BATCH_SIZE);\n    console.log(`Indexing ${totalProducts} products in ${totalBatches} batches`);\n\n    // Batch index products\n    for (let batch = 0; batch < totalBatches; batch++) {\n      const offset = batch * BATCH_SIZE;\n      // Fetch batch of active products from Postgres\n      const batchResult = await sql`\n        SELECT \n          id, \n          name, \n          description, \n          price, \n          category, \n          in_stock, \n          created_at \n        FROM products \n        WHERE active = true \n        ORDER BY created_at DESC \n        LIMIT ${BATCH_SIZE} \n        OFFSET ${offset}\n      `;\n\n      const products = batchResult.rows.map((row) => ({\n        id: row.id.toString(),\n        name: row.name,\n        description: row.description,\n        price: parseFloat(row.price),\n        category: row.category,\n        in_stock: row.in_stock,\n        created_at: row.created_at.toISOString(),\n      }));\n\n      // Add documents to Meilisearch in batch\n      try {\n        const addTask = await productIndex.addDocuments(products);\n        await meiliClient.waitForTask(addTask.taskUid);\n        console.log(`Indexed batch ${batch + 1}/${totalBatches}: ${products.length} products`);\n      } catch (batchError) {\n        console.error(`Failed to index batch ${batch + 1}:`, batchError);\n        // Retry once on failure\n        const retryTask = await productIndex.addDocuments(products);\n        await meiliClient.waitForTask(retryTask.taskUid);\n        console.log(`Retried and indexed batch ${batch + 1}`);\n      }\n    }\n\n    // Return success response with indexing stats\n    return NextResponse.json({\n      success: true,\n      total_products: totalProducts,\n      total_batches: totalBatches,\n      message: `Successfully indexed ${totalProducts} products into ${PRODUCT_INDEX}`,\n    });\n  } catch (indexError) {\n    console.error('Product indexing failed:', indexError);\n    return NextResponse.json(\n      { error: 'Failed to index products. Check logs for details.' },\n      { status: 500 }\n    );\n  }\n}\n

// Short snippet: Initialize Meilisearch in Next.js 15 Server Component\nimport { Meilisearch } from 'meilisearch';\n\nconst meili = new Meilisearch({\n  host: process.env.MEILI_HOST!,\n  apiKey: process.env.MEILI_SEARCH_KEY!, // Use read-only search key, not master key\n});\n\nexport async function ProductSearch({ query }: { query: string }) {\n  const index = meili.index('products');\n  const results = await index.search(query, { limit: 20 });\n  return ;\n}\n

// Short snippet: Update Meilisearch index settings\nconst index = meiliClient.index('products_v1');\nawait index.updateSettings({\n  searchableAttributes: ['name', 'description', 'category'],\n  filterableAttributes: ['category', 'in_stock', 'price'],\n  sortableAttributes: ['price', 'created_at'],\n  rankingRules: ['words', 'typo', 'proximity', 'attribute', 'sort', 'exactness'],\n});\n

// Short snippet: Dark launch feature flag check in search API\nconst useMeilisearch = process.env.FEATURE_FLAG_MEILI_SEARCH === 'true';\nif (useMeilisearch) {\n  return await meilisearchSearch(query, category, page);\n} else {\n  return await elasticsearchSearch(query, category, page);\n}\n