Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.19027002 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901638112870400 |
|---|---|
| author | Gintoro |
| author_facet | Gintoro |
| contents | <p>API Gateway is among the most impactful microservices design patterns for modifiability, yet prior empirical studies treat it as an indivisible unit, masking significant effectiveness variation across implementation variants. This study introduces variant-level analysis for API Gateway through a four-phase Design Science Research approach, examining 19 gateway instances across 13 open-source applications derived from an existing dataset of 110 microservices. Five gateway variants were identified through three-dimensional classification (decomposition strategy, fan-out degree, cross-cutting concern integration), with the Kruskal-Wallis test confirming significant Service Independence Metric (SIM) differences across variants (H=13.245, p=0.010, η²=0.660). Fan-out degree (SDM_out) was established as the dominant predictor of gateway SIM (Spearman r=−0.949, p<0.001), with the linear model SIM≈1.030−0.093×SDM_out explaining 77.9% of variance. Based on these findings, the Fan-out Bounded Adaptive Gateway (FBAG) model was formulated, constraining per-component fan-out to SDM_out≤2 through adaptive decomposition. Validation on all 13 applications demonstrated gateway SIM improvement of 21.5% (0.728→0.885, p<0.001, 95% CI [+0.095, +0.231]), surpassing the overall service average from the original study. The Monolithic Gateway variant (SDM_out≥5) was empirically identified as a measurable anti-pattern.</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_19027002 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Fan-out Bounded Adaptive Gateway: A Variant-Level Analysis and Enhanced Model for Improving API Gateway Modifiability in Microservices Architecture Gintoro <p>API Gateway is among the most impactful microservices design patterns for modifiability, yet prior empirical studies treat it as an indivisible unit, masking significant effectiveness variation across implementation variants. This study introduces variant-level analysis for API Gateway through a four-phase Design Science Research approach, examining 19 gateway instances across 13 open-source applications derived from an existing dataset of 110 microservices. Five gateway variants were identified through three-dimensional classification (decomposition strategy, fan-out degree, cross-cutting concern integration), with the Kruskal-Wallis test confirming significant Service Independence Metric (SIM) differences across variants (H=13.245, p=0.010, η²=0.660). Fan-out degree (SDM_out) was established as the dominant predictor of gateway SIM (Spearman r=−0.949, p<0.001), with the linear model SIM≈1.030−0.093×SDM_out explaining 77.9% of variance. Based on these findings, the Fan-out Bounded Adaptive Gateway (FBAG) model was formulated, constraining per-component fan-out to SDM_out≤2 through adaptive decomposition. Validation on all 13 applications demonstrated gateway SIM improvement of 21.5% (0.728→0.885, p<0.001, 95% CI [+0.095, +0.231]), surpassing the overall service average from the original study. The Monolithic Gateway variant (SDM_out≥5) was empirically identified as a measurable anti-pattern.</p> |
| title | Fan-out Bounded Adaptive Gateway: A Variant-Level Analysis and Enhanced Model for Improving API Gateway Modifiability in Microservices Architecture |
| url | https://doi.org/10.5281/zenodo.19027002 |