Add a benchmark to monitor performance for large dataset indexing (#9012

)

asv_bench/benchmarks/indexing.py

@@ -12,6 +12,7 @@
 nt = 500
 
 basic_indexes = {
+    "1scalar": {"x": 0},
     "1slice": {"x": slice(0, 3)},
     "1slice-1scalar": {"x": 0, "y": slice(None, None, 3)},
     "2slicess-1scalar": {"x": slice(3, -3, 3), "y": 1, "t": slice(None, -3, 3)},
@@ -74,6 +75,10 @@ def setup(self, key):
                 "x_coords": ("x", np.linspace(1.1, 2.1, nx)),
             },
         )
+        # Benchmark how indexing is slowed down by adding many scalar variable
+        # to the dataset
+        # https://github.com/pydata/xarray/pull/9003
+        self.ds_large = self.ds.merge({f"extra_var{i}": i for i in range(400)})
 
 
 class Indexing(Base):
@@ -89,6 +94,11 @@ def time_indexing_outer(self, key):
     def time_indexing_vectorized(self, key):
         self.ds.isel(**vectorized_indexes[key]).load()
 
+    @parameterized(["key"], [list(basic_indexes.keys())])
+    def time_indexing_basic_ds_large(self, key):
+        # https://github.com/pydata/xarray/pull/9003
+        self.ds_large.isel(**basic_indexes[key]).load()
+
 
 class Assignment(Base):
     @parameterized(["key"], [list(basic_indexes.keys())])